OBJECTIVE: To identify non-infectious antenatal and perinatal risk factors for cerebral palsy (CP) and its subtypes in children born at term. DESIGN: A population-based, case-control study. SETTING: The western healthcare region of Sweden. POPULATION: A population-based series of children with CP born at term during 1983-94 (n = 309) was matched with a control group (n = 618). METHODS: A total of 62 variables, maternal characteristics, and prepartal, intrapartal and postpartal variables were retrieved from obstetric records. Both univariate and multivariate analyses were performed for spastic and dyskinetic CP, and for the total CP group. MAIN OUTCOME MEASURES: Cerebral palsy (CP) and subtypes. RESULTS: Univariate analysis resulted in 26 significant risk factors for CP. Birthweight (OR 0.54, 95% CI 0.39-0.74), not living with the baby's father (OR 2.58, 95% CI 1.11-5.97), admittance to a neonatal intensive care unit (NICU) (OR 4.43, 95% CI 3.03-6.47), maternal weight at 34 weeks of gestation (OR 1.02, 95% CI 1.00-1.03) and neonatal encephalopathy (OR 69.2, 95% CI 9.36-511.89) were found to be risk factors for CP in the total CP group in our multivariate analysis. Factors during the periods before, during and after delivery were all shown to increase the risk of spastic diplegia and tetraplegia, whereas mostly factors during the period before delivery increased the risk of spastic hemiplegia, and only factors during delivery increased the risk of dyskinetic CP. Admittance to an NICU was a risk factor for all CP subtypes. CONCLUSIONS: The risk factor pattern differed by CP subtype. The presented risk factors may be useful indicators for identifying children at risk of developing CP, and helpful for targeting individuals for early intervention programmes.