The purpose of this study was to evaluate the impact of a dog safety program on the protective behaviour knowledge of children in the first year of school. This mixed methods study utilised pre- and post-testing (N = 121), observation of six sessions and individual interviews with a sub-group of children (N = 49). These interviews utilised a co-constructed narrative strategy where children were invited to assist an imaginary alien to safely navigate hypothetical safety scenarios. All schools improved their knowledge of safe dog interactions, with an overall increase in knowledge of 18%. Most children were able to apply abstract knowledge to hypothetical scenarios involving accompanied and unaccompanied dogs. Of concern, 24% of children still believed that dogs liked being patted on their heads and 16% of children had not overcome their intuitive reaction to run from a threatening dog. Whilst the program has made significant improvements to children’s knowledge of safe dog-interactions, more gains can be made. We identify important opportunities for improving dog safety programs in general. We comment on the need to consider the impact of different models of child–dog relations in terms of either similitude or difference.