SummaryThe value of routine intrapartum cardiotocography has yet to be proved despite widespread use. The confounding effects of treatment paradox and relative risk assessment are possible reasons for this.The pattern of management within 10000 consecutive deliveries was examined, using an‘independence Bayes' probability model to demonstrate changing relationships between test results, obstetric intervention and neonatal outcome.The incidence of birth asphyxia in the Prince of Wales hospital decreased by 33 percent between 1985 and 1986. This is shown to have been associated with improvements in the selection of patients for intrapartum cardiotocography and for elective caesarean section, and with an increased rate of intervention for fetal distress. The incidence of fetal distress also decreased, suggesting that factors other than intervention in response to fetal distress were partly responsible for improved perinatal outcome. A higher than expected incidence of asphyxia where delivery was by emergency caesarean section in the absence of fetal distress suggests cardiotocography can also have a detrimental effect on perinatal outcome.