Obesity prevention trials are designed to promote healthy weight. The success of these trials is often assessed using one of three metrics--means, incidence or prevalence. In this study, we point out conceptual shortcomings of these metrics and introduce an alternative that we call 'excess gain'. A mathematical demonstration using simulated data shows a scenario in which the statistical power of excess gain compares favorably with that of incidence and prevalence. Prevention of excess gain communicates an easily understood public health message that is applicable to all individuals regardless of weight status.