Methods are developed for predicting the expected precision for studies of the contribution of fish to a fishery, based upon the number of fish marked and the number of years an experiment is repeated. Studies concerned with estimating catch–release ratios, comparing catch–release ratios, and comparing distributions of catch are considered. It is suggested that releases of marked fish should be repeated for at least three or four broods, and often there is little advantage in releasing more than 50 000 marked fish per release group. Although we explicitly address studies of contribution to ocean fisheries, the methods apply directly to a broad range of studies involving marked fish, from evaluations of harvest rates on catchable-trout plants to estimates of catch–escapement ratios for Pacific salmon.Key words: precision, experimental design, number of fish to mark, number of years to release marked fish, catch–release ratios, distribution of catch