In this paper, we aimed to identify changes in the indications for and timing of caesarean section over 20 years. This involved a prospective data collection by clinical record analysis throughout the 12-month periods for 1976, 1986 and 1996. The study was carried out in the maternity unit of a large district teaching hospital. We studied all women delivered by caesarean section. Main outcome measures were rates for different indications for caesarean section for the populations served during each of the 3 years. Analysis of 1819 caesarean sections showed an increasing rate, from 6.7% in 1976 to 14.2% in 1996. The proportion of planned antepartum deliveries remained constant at 54%, with previous caesarean section given as the main indication in 1976 (55%) and 1986 (49%) and maternal request in 1996 (23%). Caesarean section for intrapartum fetal distress doubled over the study period, with little evidence of improved neonatal or long-term outcome. Caesarean section for failed labour induction and failed first- and second-stage progress all increased and for failed assisted delivery increased threefold. Caesarean section with a singleton breech presentation increased from 30% to 88% and for twin pregnancies from 13% to 47%. The rate of caesarean section for women delivered previously by section remained unaltered at 56%. The proportion of pregnancies delivered by caesarean section increased for virtually all indications. Consumer expectation has encouraged a more ready use of section, with maternal choice being the most frequent indication in 1996.