Plurality electoral systems with multi-member districts and single nontransferable votes (SNTV) allow parties to win multiple seats in district elections by nominating multiple candidates, but they also penalize a party's seat share if the number of candidates offered is ‘too many’ or ‘too few’. Given an institutional incentive to nominate the ‘correct’ number of candidates, we seek to establish empirically that the nominating behaviour of parties in such systems results from a rational calculus of strategic choice. So we develop and test an empirical theory of rational nominating behaviour applied to Japanese district elections before the 1994 electoral reform. We establish, for all possible nominating strategies, the conditions on voting outcomes required for actors to maximize benefits in the context. The efficiency of actual strategy choices for maximizing benefits is found by comparing an observed outcome from voting (the distributed benefit) with the benefit that would be expected had the party chosen its ‘best’ alternative nominating strategy instead. Empirical testing indicates that Japanese parties discriminated between available nominating strategies and made choices that maximized benefits in the context, evidence that the nominating behaviour of parties in this test environment was based on rational calculation.