Design. A group randomised controlled trial (RCT) was conducted. Intervention. Australian-rules football teams were randomly assigned to the test group (MG, custom-made mouthguards) or control group (usual mouthguard behaviour). Outcome measure. The number of head or orofacial (H/O) injuries per 1000 person–hours of playing was recorded. Results. Twenty-three Australian rules football teams were recruited of which 11 were randomised to the control group (n=111) and 12 to the test group (n=190). The majority of players, including those in the control group, wore mouthguards during games; fewer wore them every training session. Overall rate of H/O injuries was 2.7 per 1000 exposure–h, and was higher during games than training. When data were adjusted for division of play and age group there was evidence of a significant (P=0.04) protective effect of MG over control during games and training combined. The adjusted H/O injury incidence rate ratio was 0.56. Conclusions. Custom-made mouthguards provided a significant protective effect relative to usual mouthguard use during games.