STUDY OBJECTIVE--The aim was to investigate the possible association between parity, as indicated by the number of childbirths, and prevalence of urinary incontinence in an adult female population sample. DESIGN AND SETTING--A sample of 3114 women aged 30-59 years was selected at random from the population of Aarhus, Denmark, and mailed a self administered questionnaire on urinary incontinence and, among other things, parity. PARTICIPANTS--A total of 2631 questionnaires was returned (85%) with a slight but significant decrease in respondency by age. MAIN RESULTS--The 1987 urinary incontinence period prevalence was 17%. Seventy eight percent were parous, and 24% had had three or more childbirths. In women aged 30-44 years, the prevalence of urinary incontinence was found to be associated with parity and, in women aged 45 years and more, with three or more childbirths. In parous women 30-44 years of age, the prevalence of urinary incontinence increased with age at least childbirth and, in women aged 45 years and over, it increased with increasing parity but decreased with increasing age at first childbirth. In parous women, no association was found with time since last childbirth. Among clinical types of urinary incontinence, stress incontinence consistently showed the strongest associations with indicators of parity. In women aged 30-44 years, nearly two thirds of the 1987 prevalence of stress incontinence could be attributed to parity. CONCLUSIONS--These findings support the hypothesis that pregnancy and childbirth are potent causes of female urinary incontinence, so that they exert considerable impact on the level of population urinary incontinence prevalence. In the individual woman, the effect seems to be cumulative and long lasting but fades with age.