Dreaming Vol. 4, No. 2, 1994


Sleep and Dreaming Characteristics of Frequent Nightmare Subjects in a University Population

Ross Levinl,2

Self-reports of nightmare frequency were collected in seven independent nonclinical populations (n = 3433). In addition, sleep and dreaming characteristics of frequent nightmare subjects and low-nightmare controls were compared in three smaller subsamples. Reporting rates of nightmare occurrence were remarkably stable with females reporting significantly more nightmare attacks (p < .001) than males. Analysis of sleep and dreaming parameters reliably differentiated nightmare reporters from low-nightmare controls on a number of measures. Nightmare subjects rated their sleep quality as poorer (p < .05), had greater dream recall (p < .001), were more affected by their dreams (p < .001) and nightmares (p < .001), and reported more aggression in their dreams (p < .001) than controls. Thus, nightmare sufferers appear to be significantly more internally directed and more sensitive to their generalized dreaming states. The results are discussed within the context of Hartmann's (1984) nightmare theory.

KEY WORDS: sleep; dreams; nightmares.


The term nightmare often has been used to describe a vivid and terrifying nocturnal episode in which the dreamer is abruptly awakened from sleep. Nightmares are primarily REM-phase dream products which are characterized by a subjective feeling of dread or terror in which the dreamer experiences a paralysis of body movement (Kales et al., 1980). In most cases, the dreamer will have complete recall of a "long and vivid" dream product (Fisher, Byrne, Edwards, & Kuhn, 1970). REM nightmares most commonly occur towards the end of the sleep cycle and are marked by little if any autonomic arousal (Fischer et al., 1970).

While the exact etiology of nightmare occurrence is unknown, Hartmann and colleagues (Hartmann, 1984, 1991; Hartmann, Russ, Oldfield, Sivan, & Cooper, 1987; Hartmann, Russ, van der Kolk, Falke, & Oldfield, 1981) have suggested that frequent nightmares reflect a structural impairment in "ego boundaries" which fail to protect the dreamer from perceived threats both during sleep and in their waking states. As a result, nightmare sufferers have been characterized as unusually open and vulnerable people with a "surprising lack of the usual defense mechanisms" (Hartmann et al., 1981, p. 796).

Consistent with this formulation, individuals who experience frequent, lifelong nightmares have been repeatedly shown to demonstrate greater psychopathology than matched controls without nightmares (Hartmann et al., 1981, 1987; Hersen, 1972; Kales et al., 1980; Kramer, Schoen & Kinney, 1984; Levin, 1989, 1990, 1993a; Levin & Raulin, 1988, 1991; Levin, Galin & Zywiak, 1991; van der Kolk, Blitz, Sherry, & Hartmann, 1984). Based upon similar etiological considerations regarding ego boundary impairment and schizophrenia (Blatt & Wild, 1976), it has been hypothesized that frequency nightmare sufferers may constitute a particular high-risk group for psychosis-proneness (Hartmann, 1984; Levin, 1993a; Levin & Raulin, 1988, 1991). (See Levin & Fireman (1993) for a more thorough review of this literature.) Given the relationship between nightmare frequency and ego boundary impairment and concomitant thought disorder (Levin, 1990; Levin, Galin, & Zywiak, 1991), an examination of the general sleep and dreaming patterns of frequent nightmare reporters could help in understanding the factors which predispose these individuals to their attacks.

While the parameters of the actual nightmare dream have received some attention in the literature (Fischer et al., 1970; Hartmann et al., 1987; Kramer, Schoen & Kinney, 1984; Levin & Fireman, 1993), there has been little attempt to compare empirically overall sleep and dreaming patterns of nightmare subjects and controls, particularly utilizing large sample populations. Such an undertaking would seem to be of considerable interest given the hypothesized connection between frequent nightmare occurrence and both schizotypal personality (Hartmann et al., 1981; Levin & Raulin, 1991) and increased accessibility to the creative process (Hartmann, 1984; Levin, Galin, & Zywiak, 1991).

The purpose of the present investigation then was to provide normative data on the sleep and dreaming characteristics of individuals who report experiencing frequent nightmares in a large, non-clinical population. Because time limitations prevented a more in-depth investigation of nightmare frequency, data was obtained from multiple independent samples (seven introductory psychology classes), thus providing some check on the reliability of self-reports of nightmares. In addition, descriptive data on a number of salient dimensions related to sleep and dreaming were collected on three random independent subgroups of this population to assess whether other variables differentiated nightmare subjects from ordinary dreamers. The investigation of sex differences was of particular interest in this study given previous evidence that women are more likely to report having nightmares and may experience these attacks with greater fearfulness than men (Cuddy & Belicki, 1992; Hartmann, 1984; Levin, 1989).


Subjects and Procedure

3433 college students (1509 males and 1924 females) enrolled in introductory psychology at the State University of New York at Buffalo over a period of four years completed a one-page nightmare frequency checklist as part of a large packet of questionnaires routinely administered at the beginning of each academic semester. The vast majority (n = 2885) were first year students (84%) and the mean age of the overall sample was 18.7. As psychology is a required class for most students, the defining characteristics of this large sample may be estimated to approximate the overall demographic breakdown of the entering university population. Thus, approximately 12% of this group may be defined as minority (African-Americans, Hispanics, Orientals, American Indians primarily) with most students (98%) being New York State residents, 50% from the Western New York region. About 78% are from urban backgrounds, with the rest emanating from rural environments. Given that SUNY-Buffalo is a public institution, the SES of most students ranges from working to upper middle-class with an emphasis on the former.

Consistent with previous research (Hersen, 1971; Levin & Raulin, 1991), nightmares were operationally defined as "a scary dream that awakens you from sleep." Subjects checked one of six possible frequency categories: not at all, very infrequently (about l x /year), few times a year, once a month, once a week, more than once a week. Of these, 306 (9%) completed an in-depth Sleep and Dreaming Survey developed by the author in three independent sessions over a three year time span. The samples consisted of 58 (all female), 169 (65 male and 104 female) and 79 (37 male, 42 female) subjects respectively. In samples one and three, equal numbers of high nightmare (once a month or greater) and low nightmare (very infrequently) subjects were those who were preselected for participation in another nightmare study by the author from the larger pool for comparison. In sample two (n = 169), all subjects who participated in the initial nightmare frequency data collection phase were given the more extensive questionnaire. The mean age of each of these three subsamples were 18.8, 19.1, and 19.2.

The scale consists of 34 items designed to assess descriptive and quantitative data on subjects' quality of sleep (i.e., "How restful is your sleep?"), dreaming and nightmare parameters (i.e., "How vivid would you rate your dreams?"; "How affected are you by your nightmares the following day?"), as well as the presence of other sleep disorders (excessive thrashing, sleepwalking, insomnia, difficulty breathing, talking, seizures, bedwetting). Most of the reported results are based on 7-point Likert-scale ratings with higher numbers representing a quantitative increase.

An extreme contrasted groups experimental design was utilized in order to best provide clear cut comparisons between subjects who reported nightmares and those who did not on the pertinent sleep and dreaming variables and t-tests were conducted on pertinent variables to assess for any group mean differences. In addition, 2 x 2 ANOVA's with sex and nightmare condition as the independent variables were conducted on the third sample to assess for the presence of interactions between these conditions.


83% of all subjects (n = 2837) reported experiencing at least one nightmare within the past year with 15% (n = 515) reporting nightmares at least once a month, a criterion often used for defining frequent nightmare sufferers (Hersen, 1971). Moreover, 5% (n = 159) of the sample met the more stringent criterion of one nightmare or more per week. There were no significant differences in reporting rates across the seven independent samples comprising the overall population nor were there any significant variations in demographic data across these semesters.

Of special interest, females were about 50% more likely to report frequent nightmare attacks than males. Chi-square analyses on these sex differences were highly significant, x2 = 21.99, p < .001. A complete breakdown of nightmare frequency by gender is presented in Table 1.

An investigation of the Sleep and Dreaming Survey reveals consistent differences between the nightmare group and the controls across all three subsamples. A summary of the major findings are presented in Table 2.

Sample I

In the first study, 19 analyses were conducted with 10 reaching statistical significance and all means in the predicted direction of increased salience for the nightmare group. As seen in Table 2, high nightmare subjects rated the quality of their sleep as poorer than the controls, and had a higher incidence of nocturnal awakenings.



In addition, nightmare subjects differed considerably from the controls on a number of dreaming dimensions. Significant differences were found in dream recall frequency, degree of being affected by dreams, and likelihood of recurring dreams. Furthermore, there were considerable differences between groups regarding dream content. Frequent nightmare subjects reported more aggression in their dreams and rated their aggression as more intense than the controls. A content analysis of common emotions in dreams revealed that controls were twice as likely to have 'happy dreams' and that the nightmare group often used negative adjectives to describe their most common dreaming emotions. Thus, such descriptive terms as helplessness, paranoid, hurt, shock, guilt, jealousy, crying, yelling, frustration, death and restlessness were found only in the records of nightmare subjects. Not surprisingly, nightmare subjects were also significantly more affected by their nightmare dreams and were generally more likely to report additional sleep disturbances than controls.

Sample 2

The above results were generally replicated in the second independent sampling utilizing a larger population. As in the first group, 19 analyses were run with 9 reaching statistical significance. In fact, with the exception of incidence of other sleep disorders, all other findings were duplicated. However, as in the first sample, when nightmare subjects endorsed additional sleep disorders, they were far more likely to attribute greater severity to their condition than the controls. In addition, the nightmare sufferers in the second sample rated their bad dreams as far more intense than the controls, a finding which was suggested in the first sample but short of significance.

However, a number of interesting sex differences emerged in this sample. While gender was not a significant factor in overall dream recall, t(167) = -.55, women were far more sensitive to their dreaming states. Thus, females rated their non-nightmare dreams as being more vivid, t(167) = 2.72, p < .007, more meaningful, t(166) = 2.71, p < .008, and having a greater waking effect, t(167) = 3.27, p < .001, than did the male subjects. These findings were even more powerfully demonstrated in the ratings of actual nightmares. Females experienced these dreams with more realism, t(141) = 2.90, p < .004, and vividness, t(141) = 2.55, p < .01, and were significantly greater affected by their nightmares, t(141) = 3.96, p < .001, than males.

Sample 3

The findings of the first two samples were replicated and extended in this group. For the 17 variables analyzed, there were 13 significant effects for nightmare condition and three for sex, most at a powerful level of statistical significance. Furthermore, 2 x 2 AVOMA's on these data revealed two interactions.

Once again, nightmare subjects rated their sleep quality as poorer than controls and reported significantly greater dream recall. Furthermore, nightmare subjects reported that their dreams were more vivid, contained greater movement and had more personal meaning than low-nightmare controls. The non-nightmare dreams of the experimental group also contained more aggressive content and a greater intensity of aggression than the control group. As might be expected, these subjects also reported their nightmares to be significantly more vivid, realistic, and intense than the low-nightmare group.

As in sample 2, there were a number of sex differences, with women reporting a greater prevalence of aggressive content in their non-nightmare dreams, t(74) = 3.24, p < .004, a greater incidence of nightmares, t(74) = 2.31, p < .05, and considerably greater distress over these nightmares, t(74) = 2.67, p < .04, than men. Other differences reported in the previous sample did not reach statistical significance, perhaps due to the large variation in sample size. There were significant group by sex interactions for dream affectedness, F(1, 72) = 3.89, p < .05, and nightmare distress, F(1, 62) = 6.43, p < .01. A closer examination of these variables as presented in Figure 1 reveal considerable consistency. On both variables, females evidenced greater reactivity to their dream products as a function of their nightmare incidence, while males responded about equally to their internal states, irrespective of their nightmare frequency. Thus, gender differences on these variables appear to be specific only to the high-nightmare group.




The findings of the present study confirm and extend previous formulations regarding normative data on nightmare subjects (Belicki & Belicki, 1982; Hartmann et al., 1987; Kramer et al., 1984) and strongly suggest that there are sex differences in the incidence of nightmare reporting. The finding that 15% of the sampled group report monthly nightmares falls almost exactly between the figures reported by Feldman & Hersen (1967) and Belicki & Belicki (1982) although the number of reported weekly nightmare sufferers (5%) is identical with the former investigators. In addition, the finding that 83% of the sample reported at least one nightmare within the past year underscores the ubiquity of the occasional nightmare experience, at least in young adults. Clearly, survey data which assessed nightmare characteristics in a broader age and demographic range would be of considerable interest and given the restricted sample, caution should be utilized in generalizing from these data.

The consistency with which females report a greater incidence of nightmares remains an interesting yet puzzling finding and has been reported by other investigators (Cerenovsky, 1984; Feldman & Hersen, 1967; Hersen, 1971). Given the reliance on self-report measures, it is difficult to determine whether females are actually more susceptible to nightmare attacks or are simply more open to reporting such information (Hartmann, 1984). In this sense, these data are consistent with reviews of sex differences in personality which conclude that women report both a greater quantity and intensity of fears than men on fear surveys (Buss, 1988). In addition, recent work by Cuddy and Belicki (1992) suggest that gender differences in reporting rates greatly diminish when prior history of sexual abuse is controlled for. While a prospective dream diary study would be invaluable in validating these retrospective screening instruments, the data from the second and third subsamples suggest that these higher reporting rates may be a component of females' heightened overall sensitivity to their generalized dreaming states. Given that the degree of perceived distress may be a mediating factor in the level of pathology associated with nightmare incidence (Belicki, 1992), the present data, which indicates that female nightmare sufferers exhibit the greatest level of sensitivity to their internal experiences, suggest that these individuals may be particularly likely to experience nightmares with greater clarity, and thus with increased anxiety and fearfulness and that these effects may become more readily generalized to waking experiences (e.g., psychological distress and heightened psychopathology). Thus, these findings may help to clarify previous research (Levin, 1989; Levin & Raulin, 1991) suggesting that female nightmare subjects demonstrate greater waking psychopathology than male nightmare subjects. However, given the reliance on retrospective estimates of nightmare frequency in the present study, these conclusions should be regarded with caution and additional research utilizing dream logs on the interaction between sex and nightmare experience is warranted.

Perhaps the most salient finding across all samples was the degree to which the nightmare subjects reported greater overall attentiveness to their dreamlife in general. Consistent with Hartmann et al. (1987), nightmare subjects were far more likely to remember their dreams than controls and were considerably more affected by them the following day. These similarities are strengthened further given the different methods of data collection in the two studies (clinical interview vs. self-report survey), as well as differences in the samples utilized (life-long nightmare sufferers vs non-clinical college students).

Taken together, frequent nightmare subjects appear to be more internally-directed and therefore, more responsive to their dreaming states in general than the low-nightmare controls. These data are consistent with the work of Kramer and colleagues (1990; Kramer, Schoen & Kinney, 1984, 1987) who have shown that disturbed dreamers are more responsive, both emotionally and physiologically, to their dreams than controls. The present investigation confirms these findings with a considerably augmented sample and are consistent with Kramer's (1990) contention that dream disturbances reflect an integrative failure for emotionally arousing internal events. Heightened sensitivity to dream content, which under even optimal conditions appears to be predominated by negatively-toned affect (Hall & Van de Castle, 1966), could also explain the greater incidence and intensity of experienced aggression in the dreams of these subjects.

As suggested above, these findings can also be explained within the context of Hartmann's (1984) nightmare theory. According to Hartmann, nightmare sufferers are characterized by thin or permeable "boundaries in the mind" which render the dreamer highly vulnerable to internal intrusions from optimally-repressed material which are then in turn experienced as threats. Thus, it may not be the intensity of the cognitions per se which lead to nightmares but rather, a breakdown in the self-protective mechanism which gives rise to the experience of the attack. Because of these deficits, nightmare sufferers may become overly susceptible to experiencing psychic dissolution in the face of a perceived threat to the stability of their psychological sense of self (Levin & Hurvich, in press). Similar conclusions have been offered by Cooper & Hartmann (1986) regarding the lack of overt hostility in the TAT protocols of nightmare sufferers.

In this sense, it is interesting to note that frequent nightmare sufferers have been described as being unusually open and in many cases, to have heightened accessibility to primary process thought (i.e., creativity) (Hartmann, 1984; Hartmann et al., 1987; Levin, Gafin,, & Zywiak, 1991). It thus appears that the dream may hold far greater meaning in the mental lives of individuals with frequent nightmares than in those who do not have such experiences. The significant interactions found in the third sample supports this formulation. While this may be explained as a natural concern and focused attention on a problem area, it is also possible that this increased introspection represents an over compensatory mechanism to establish greater self-constancy and regain a sense of security. Consistent with this, preliminary Rorschach data from another study (Levin, 1993b) indicates that nightmare subjects are more likely to evidence an introversive cognitive style than matched controls. Over-reliance on such a coping style, which places primary emphasis on internal fantasy in problem-solving has also been documented in studies of schizotypal personality (Exner, 1986).

In accordance with directions suggested by the present data, future research should examine whether the degree of heightened dream introspection generalizes to a higher level of waking self-awareness. Should this be the case, these data could have welcome implications in assessing suitability for insight-oriented therapy as well as possibly providing a priori prognostic information regarding amenability to treatment. Indeed, there is evidence to suggest that heightened focal awareness on dreams can lead to symptomatic relief (Fiss, 1986) as well as facilitate amenability to psychological treatment (Cartwright, Tipton & Wicklund, 1980). Last, it is suggested that future research focus more closely on the actual content of the nightmares as well as of the non-nightmare dreams of the nightmare sufferer. In this regard, Kramer (1990) has suggested an innovative framework for investigating possible information-processing functions of dreams by examining various intra-night dreaming patterns in frequent nightmare populations. In addition, it should also be possible to investigate whether the dream products and nightmares of the frequent sufferer reflect the lack of boundary integrity which Hartmann (1984) hypothesizes. In this way, systematic empirical investigation of these dreams could provide a more direct test for underlying etiological theories of nightmares.


Belicki, D., & Belicki, K. (1982). Nightmares in a university population. Sleep Research, 11, 116.

Belicki, K. (1992). The relationship of nightmare frequency to nightmare suffering with implications for treatment and research. Dreaming, 2, 143-148.

Blatt, S. J., & Wild, C. M. (1976). Schizophrenia: A developmental analysis. New York: Academic Press.

Buss, A. (1988). Personality: Evolutionary heritage and human distinctiveness. Hillsdale, N.J.: Lawrence Erlbaum.

Cartwright, R., Tipton, L., & Wickland, J. (1980). Focusing on dreams. Archives of General Psychiatry, 37, 275-277.

Cernovsky, Z. Z. (1984). Life stress measures and reported frequency of sleep disorders. Perceptual and Motor Skills, 58, 39-49.

Cooper, S. H. & Hartmann, E. (1986). Hostility levels of lifetime nightmare sufferers A test of a clinical hypothesis. Psychoanalytic Psychology, 3, 373-377.

Cuddy, M. A., & Belicki, K. (1992). Nightmare frequency and related sleep disturbance as indicators of a history of sexual abuse. Dreaming, 2, 15-22.

Exner, J. E. (1986). Some Rorschach data comparing schizophrenics with borderline and schizotypal personality disorders. Journal of Personality Assessment, 50, 455-471.

Feldman, M. J., & Hersen, M. (1967). Attitudes towards death in nightmare subjects. Journal of Abnormal Psychology, 72, 421-425,

Fisher, C., Byrne, J., Edwards, A., & Kahn, E. (1970). A psychophysiological study of nightmares. Journal of the American Psychoanalytic Association, 18, 747-782.

Fiss, H. (1986). An empirical foundation for a self psychology of dreaming. Journal of Mind and Behavior, 7, 161-192.

Hall, C., & Van de Castle, R. (1966). The content analysis of dreams. New York: Appleton.

Hartmann, E. (1984). The nightmare: The psychology and biology of terrifying dreams. New York: Basic Books.

Hartmann, E. (1991). Boundaries in the mind: A new psychology of personality. New York: Basic Books.

Hartmann, E., Russ, D., Van der Kolk, B., Falke, R., & Oldfield, M. (1991). A preliminary study of the personality of the nightmare sufferer: Relationship to schizophrenia and creativity? American Journal of Psychiatry, 138, 794-797,

Hartmann, E., Russ, D., Oldfield, M., Sivan, I., & Cooper, S. (1987). Who has nightmares? The personality of the lifelong nightmare sufferer. Archives of General Psychiatry, 44, 49-56.

Hersen, M. (1971). Personality correlates on nightmare sufferers. Journal of Nervous and Mental Disease, 153, 27-31.

Hersen, M. (1972). Nightmare behavior: a review. Psychological Bulletin, 78, 37-48.

Kales, A., Kales, J. D., Soldatos, C. R., Caldwell, A. B., Charney, D. S., & Martin, E. D. (1980). Nightmares: Clinical characteristics and personality patterns. American Journal of Psychiatry, 137, 1197-1201.

Kramer, M. (1990). Nightmares (dream disturbances) in posttraumatic stress disorder. In R. R. Botzin, J. F. Kihlstrom, & D. L. Schacter (Eds.), Sleep and cognition. APA: Washington.

Kramer, M., Schoen, L. S., & Kinney, L. (1984). Psychological and behavioral features of disturbed dreamers. Psychiatric Journal of the University of Ottawa, 9, 102-106.

Kramer, M., Schoen, L. S., & Kinney, L. (1987). Nightmares in Vietnam veterans. Journal of the American Academy of Psychoanalysis, 15, 67-81.

Levin, R. (1989). Relations among nightmare frequency and ego strength, death anxiety, and sex of college students. Perceptual and Motor Skills, 69, 1107-1113.

Levin, R. (1990). Ego boundary impairment and thought disorder in frequent nightmare subjects, Psychoanalytic Psychology, 7, 529-544.

Levin, R. (1993a). Nightmares and psychosis-proneness. Manuscript submitted for publication.

Levin, R. (1993b). Rorschach correlates of frequent nightmare sufferers. Manuscript submitted for publication.

Levin, R., & Fireman, G. (1993). Psychoanalytic studies on the experience of schizophrenia. In J. Masling & B. Bornstein (Eds.), Empirical studies of psychoanalytic theories, Vol. 4. (pp. 139-189), Washington, D.C.: APA Press.

Levin, R. & Hurvich, M. (in press). Nightmares and annihilation anxiety. Psychoanalytic Psychology.

Levin, R., Galin, J., & Zywiak, W. (1991). Nightmares, boundaries, and creativity. Dreaming, 1, 63-74.

Levin, R., & Raulin, M. L. (1988, August). Dreams, nightmares, and schizotypal personality organization. Paper presented at the meeting of the American Psychological Association, Atlanta, GA.

Levin, R., & Raulin, M. L. (1991). Preliminary evidence for the proposed relationship between frequent nightmares and schizotypal symptomatology. Journal of Personality Disorders, 5, 8-14.

Van der Kolk, B., Blitz, R., Burr, W., Sherry, S., & Hartmann, E. (1984). Nightmares and trauma: A comparison of nightmares after combat with life-long nightmares in veterans. American Journal of Psychiatry, 141, 187-190.

1Yeshiva University, New York.

2'Requests for reprints should be addressed to Ross Levin, Ph.D., Ferkauf Graduate School of Psychology, Albert Einstein College of Medicine Campus, 1300 Morris Park Avenue, Bronx, N.Y. 10461.

  Copyright 2003 Association for the Study of Dreams. All Rights Reserved.