The feasibility of using genetic stock identification to analyse seasonal changes in stock compositions of Atlantic salmon catches in the Baltic Sea was examined. The analysis employed seven variable allozyme loci from most of the potentially contributing stocks (16) from Finland and Sweden. Catch samples were collected from Finnish salmon fisheries in the eastern Bothnian Sea during the 1992 fishing season. Simulation studies were used to evaluate the feasibility of identifying Baltic salmon stocks with allozyme data. Special attention was paid to analysing the wild production of salmon stocks. Clear seasonal differences in stock composition were found. The estimates were compared with smolt production and Carlin-tag data. The proportions of the Neva and Oulujoki river stocks could be estimated as individual stocks, whereas the contributions of the remaining stocks were estimated as four composite stock groups. One of the groups consisted of wild stocks from the rivers Kalixälven and Simojoki. Identification of this group, which could be used as an index of wild production in the catches, requires catch sample sizes >300 salmon if <15% error is required.