We test the theoretical prediction that profit sharing reduces worker separations and by doing so increases the incidence of training. Using individual level UK data, we confirm that profit sharing is a robust determinant of lower separation rates and of greater training incidence. Critically, we cannot confirm the predicted link between separations and training. Instead, the evidence supports alternative theories suggesting a direct link between profit sharing and training. Our results suggest that profit sharing changes employer-worker relations in a way that leads to greater formal and informal investment in worker skills but that this is independent of its influence on reducing separations.