Atlantic salmon (Salmo salar) males mature as either tiny precocious parr before seaward migration, or as older and larger anadromous males. Anadromous males dominate the spawning redds and aggressively defend females against parr intrusions. Parr gain fertilizations by sneaking in to ejaculate while anadromous males and females spawn. Such differences in mating advantage generate asymmetries in risk of sperm competition between the male strategies. Theoretical sperm competition models predict that males typically mating in disfavoured roles (here, the parr strategy) should be selected to offset this disadvantage by investing more into spermatogenesis to achieve fertilization success. First, we present a theoretical model which analyses gametic expenditure for salmon parr and anadromous male reproductive strategies. We then use the natural variance in mating pattern within this species to compare empirically how males invest in spermatogenesis. A range of reproductive traits were measured for both male strategies. Absolutely, anadromous males have larger testes and produce greater numbers of sperm than parr males. However, results show that parr invest relatively more heavily into total spermatogenesis, and have a larger gonosomatic index than anadromous males. Relative to body size, parr produced greater numbers of sperm and volumes of stripped ejaculate. There was no difference in sperm length between the two male strategies. However, more sperm were motile in parr ejaculates, and these sperm lived longer than anadromous male sperm. Our findings may explain how male parr, under elevated risks of sperm competition and occupying a disfavoured mating role (parr weigh only 0.15% of the average body mass of anadromous males) achieve disproportionately high fertilization success.