The purpose of this study was to assess a model of risk for cerebral palsy (CP) in preschool children. The research was structured as a retrospective case-control study in which cases (n - 112) were children with CP while controls (n = 153) were either friend referrals or populationbased. Data were collected via structured telephone interviews. While significant associations with CP were found on univariate analyses for a number of variables, only four variables–birth weight/gestational age, 5-min APGAR, medication for miscarriage, and urinary tract infection during pregnancy–were found predictive in multivariate analysis. These variables accounted for 36% of the log-likelihood explained by the model. © 1993 John Wiley & Sons, Inc.