OBJECTIVE: Pre-post-randomized design evaluated The Blue Dog, a dog safety software program. METHODS: 76 children aged 3.5-6 years completed 3 tasks to evaluate dog safety pre- and postintervention: (a) pictures (recognition of safe/risky behavior), (b) dollhouse (recall of safe behavior via simulated dollhouse scenarios), and (c) live dog (actual behavior with unfamiliar live dog). Following preintervention evaluation, children were randomly assigned to dog or fire safety conditions, each involving 3 weeks of home computer software use. RESULTS: Children using Blue Dog had greater change in recognition of risky dog situations than children learning fire safety. No between-group differences emerged in recall (dollhouse) or engagement (live-dog) in risky behavior. Families enjoyed using the software. CONCLUSIONS: Blue Dog taught children knowledge about safe engagement with dogs, but did not influence recall or implementation of safe behaviors. Dog bites represent a significant pediatric injury concern and continued development of effective interventions is needed.