Mate choice by males has been recognized at least since Darwin's time, but its phylogenetic distribution and effect on the evolution of female phenotypes remain poorly known. Moreover, the relative importance of factors thought to underlie the evolution of male mate choice (especially parental investment and mate quality variance) is still unresolved. Here I synthesize the empirical evidence and theory pertaining to the evolution of male mate choice and sex role reversal in insects, and examine the potential for male mating preferences to generate sexual selection on female phenotypes. Although male mate choice has received relatively little empirical study, the available evidence suggests that it is widespread among insects (and other animals). In addition to 'precopulatory' male mate choice, some insects exhibit 'cryptic' male mate choice, varying the amount of resources allocated to mating on the basis of female mate quality. As predicted by theory, the most commonly observed male mating preferences are those that tend to maximize a male's expected fertilization success from each mating. Such preferences tend to favour female phenotypes associated with high fecundity or reduced sperm competition intensity. Among insect species there is wide variation in mechanisms used by males to assess female mate quality, some of which (e.g. probing, antennating or repeatedly mounting the female) may be difficult to distinguish from copulatory courtship. According to theory, selection for male choosiness is an increasing function of mate quality variance and those reproductive costs that reduce, with each mating, the number of subsequent matings that a male can perform ('mating investment') Conversely, choosiness is constrained by the costs of mate search and assessment, in combination with the accuracy of assessment of potential mates and of the distribution of mate qualities. Stronger selection for male choosiness may also be expected in systems where female fitness increases with each copulation than in systems where female fitness peaks at a small number of matings. This theoretical framework is consistent with most of the empirical evidence. Furthermore, a variety of observed male mating preferences have the potential to exert sexual selection on female phenotypes. However, because male insects typically choose females based on phenotypic indicators of fecundity such as body size, and these are usually amenable to direct visual or tactile assessment, male mate choice often tends to reinforce stronger vectors of fecundity or viability selection, and seldom results in the evolution of female display traits. Research on orthopterans has shown that complete sex role reversal (i.e. males choosy, females competitive) can occur when male parental investment limits female fecundity and reduces the potential rate of reproduction of males sufficiently to produce a female-biased operational sex ratio. By contrast, many systems exhibiting partial sex role reversal (i.e. males choosy and competitive) are not associated with elevated levels of male parental investment, reduced male reproductive rates, or reduced male bias in the operational sex ratio. Instead, large female mate quality variance resulting from factors such as strong last-male sperm precedence or large variance in female fecundity may select for both male choosiness and competitiveness in such systems. Thus, partial and complete sex role reversal do not merely represent different points along a continuum of increasing male parental investment, but may evolve via different evolutionary pathways.