Parents of 171 children coming to the Yale-New Haven Hospital Primary Care Center for their 6-month checkup were randomized into an intervention group (n = 85) and a control group (n = 86). Parents in the intervention group received a three-part individualized course in child safety that required active parental participation. Parts 1, 2, and 3 were given at the 6-month, 9-month, and 12-month well-child visits, respectively. Parents in the control group received routine safety education as provided at well-child visits. The educational phase of the study was completed by 129 families, 65 in the intervention group and 64 in the control group. Safety knowledge, number of hazards in the home, and reported accidents were assessed by a "blinded" community health worker approximately 1 month after the 12-month well-child visit. A total of 109 home visits were made, 55 for the intervention group and 54 for the control group. Parental safety knowledge was assessed based upon pictorial hazard recognition. Of 13 possible hazards, the mean number of hazards recognized by the intervention group parents was 9.4 (n = 55) v 8.4 (n = 50) by the control group parents (t = 2.1, P less than .05, two-tailed). A hazard score was determined for each family based on nine possible hazards observed at the home visit. The mean hazard score for the intervention group was 2.4 (n = 55 v 3.0 (n = 54) for the control group (t = 2.4, P less than .02, two-tailed). Parentally reported accidents and accidents reported in hospital records were similar for both groups. Results of this study suggest that age-appropriate safety education that is repetitive and individualized and that requires active parental participation results in an increase in parental knowledge and an improvement in certain safety practices.