Introduction Sleep disturbances during pregnancy affect delivery outcomes, including shorter pregnancies and longer labour (Okun, 2009). Studies also reveal associations between recall of negative dreams and shorter labour (Mancuso et al., 2008; Winget and Kapp, 1972). However, prospective longitudinal studies assessing the value of pregnancy dreams in predicting delivery outcomes are lacking. Whether dream measures constitute better predictors of delivery outcomes than do sleep measures also remains to be determined. This 2-stage longitudinal study assessed whether prospective dream measures during pregnancy (Time-1; T1) predict delivery outcomes (Time-2; T2) controlling for pregnancy, sleep and psychosocial risk factors. Materials and methods Fifty-two 3rd trimester pregnant women ( M = 29.78 ± 3.62 wks of gestation) aged 18–37 years ( M = 28.48 ± 4.06 yrs) and not planning to undergo caesarean sections were followed-up post-delivery. At T1, pregnant women completed demographic and psychological questionnaires and a 14-day home log to assess sleep/dream characteristics. Gestational length and labour duration were assessed at T2. Deliveries occurred at 36–42 weeks of gestation ( M = 39.29 ± 1.46); labour lasted 1–40 h ( M = 13.62 ± 8.99). Four women had unplanned caesareans; their labour duration did not differ from women who delivered vaginally ( M = 13.40 ± 9.28 vs. 16.25 ± 0.03; p = 0.60). Two sets of hierarchical multiple regression analyses were run to predict delivery outcomes from pregnancy dream variables (#dreams, positive–negative dreams ratio, dream clarity, dream impact), controlling for sleep (duration, quality, bedtime/rise time, night awakenings), demographic (age, employment status, family income, education) and psychosocial (mood/concerns, stress, history of emotional problems) characteristics. All demographic variables, prenatal psychosocial factors and sleep variables associated with outcomes at p Results More negative dreams during pregnancy predicted shorter gestation (Beta = −0.61, p = 0.00002); no adjustment was required. Controlling for age, education, stress and sleep duration, later bedtime/rise time predicted longer labour (Beta = 0.28, p = 0.04). Conclusion The study provides new evidence that pregnant women who report more negative dreams are at greater risk for shorter pregnancies. Results are also consistent with the possibility that chronotype mediates relationships between sleep and labour duration. Acknowledgement Research was supported by the Canadian Institutes of Health Research.