We present an empirical test of the ‘Ghiselin—Reiss small-male hypothesis’ for the evolution of sexual size dimorphism (SSD). In mating systems dominated by scramble competition, where male reproductive success is a function of encounter rate with females, small males may be favoured when food is limiting because they require lower absolute amounts of food. Given a trade-off between time and energy devoted to foraging and to mate acquisition, small males should be able to devote more time to the latter. If at the same time larger females are favoured, this mechanism will contribute to the evolution of SSD and may be the major determinant of the female-biased SSDs that characterize most animal taxa. We tested this hypothesis using the water strider,Aquarius remigis (Heteroptera: Gerridae), a scramble competitor which mates many times over a prolonged mating season and which shows female-biased SSD. Laboratory experiments demonstrated that foraging success and giving up times (GUTs) are lower for males than for females during the reproductive season and that male water striders flexibly alter their time budgets under conditions of energy limitation. Controlled feeding experiments showed that male and female longevity, female fecundity and male mating success are positively related to food availability. As predicted, male body size is negatively correlated with several indices of male fitness (longevity, number of mating attempts and mating success), while female body size is positively correlated with longevity. These results are consistent with the hypothesis that scramble competition for mates favours small males in this species and provides empirical support for the Ghiselin—Reiss small-male hypothesis.