Prospective information gathered through the course of pregnancy, perinatal measurements, and retrospective data collected postnatally were used to investigate the changing patterns and effects of caffeine use of 286 women participating in the Ottawa Prenatal Prospective Study. Data were collected on maternal use of tea, coffee, caffeinated soft-drinks, chocolate bars and drinks and caffeinated medication. The volume and analysed caffeine concentration of 53 samples of coffee and tea, prepared by subjects as they usually consumed it, were used to examine the predictive potential of the women's subjective description of the beverages. Self-reports of volume and beverage strength were found to be valid predictors; the method of coffee preparation held little predictive power. An algorithm for estimating caffeine intake retrospectively over time was developed. During pregnancy most women continued to consume caffeine but usually at lower intake levels. After pregnancy, caffeine consumption tended to persist at reduced levels for several months and then returned to prepregnancy patterns. Maternal caffeine intake of more than 300 mg daily during pregnancy was associated with lowered birth weight and smaller head circumference of the infant after accounting for maternal nicotine use. No relationship was apparent between maternal caffeine use and the incidence of caesarian sections, breech births, miscarriages or premature births.