The objective of the study was to identify the origin (s) of the association between cerebral palsy (CP) and birth defects in the absence of cerebral birth defects. Data from the 1980 to 1994 Western Australian birth cohorts (355 659 neonatal survivors) were linked to the Cerebral Palsy Register (941 links) and the Birth Defects Registry (17 070 links). Associations between CP (congenital or acquired) and birth defects (cerebral or exclusively non-cerebral) were estimated. The origin of the association between non-cerebral defects and acquired CP was investigated with an observational study, and the origin of the association between non-cerebral defects and congenital CP was investigated with a blinded casecontrol study of births with non-cerebral defects with or without CP. With non-cerebral defects, the odds ratio for CP was 4.8 (95% CI 3.1-7.4) if acquired and 4.7 (3.9-5.7) if congenital. For acquired CP, the association arose primarily as a result of cardiac defects. For congenital CP, the association arose partly from ascertainment bias and partly from defects known to be associated with cerebral defects (but not identified in these data). However, a significant portion remained unexplained. The presence of non-cerebral defects should heighten clinical alertness to the possibility of CP and of cerebral birth defects.