The distribution of the duration of labor during the day was analyzed. Material: A one-year material consisting of 1,881 deliveries at the Department of Obstetrics, Regional Hospital of Trondheim, Norway. Only singleton pregnancies with spontaneous onset of contractions and intended vaginal delivery were included. Duration of labor was routinely recorded by the midwife. Methods: Stepwise multiple regression. Logarithmic transformation of the dependent variable (duration measured in hours) was applied. Parity (0/1+) and hour of birth were the independent variables. Results: The variable indicating delivery during the night (0 to 08) contributed significantly to explain the variation in the duration of labor, in terms of a decrease in the median duration. The effect is substantial, median duration of labor 1.2 h shorter than for deliveries during the day. The variable indicating delivery during the evening (16-24) did not reach the required level of significance (0.05). This reflects a clustering of cases with longer duration of labor, unfavourable progress and less good prognosis during the latter half of the day, which has practical implications for the obstetrical staff. This may explain the similar circadian pattern in perinatal mortality which has been repeatedly documented, the etiology of which has remained enigmatic.