Theory predicts that costly sexual traits should be reduced when individuals are in poor condition (i.e. traits should exhibit condition‐dependent expression). It is therefore widely expected that male ejaculate traits, such as sperm and seminal fluid, will exhibit reduced quantity and quality when dietary nutrients are limited. However, reported patterns of ejaculate condition dependence are highly variable, and there has been no comprehensive synthesis of underlying sources of such variation in condition‐dependent responses. In particular, it remains unclear whether all ejaculate traits are equally sensitive to nutrient intake, and whether such traits are particularly sensitive to certain dietary nutrients, respond more strongly to nutrients during specific life stages, or respond more strongly in some taxonomic groups. We systematically reviewed these potential sources of variation through a meta‐analysis across 50 species of arthropods and vertebrates (from 71 papers and 348 effect sizes). We found that overall, ejaculate traits are moderately reduced when dietary nutrients are limited, but we also detected substantial variation in responses. Seminal fluid quantity was strongly and consistently condition dependent, while sperm quantity was moderately condition dependent. By contrast, aspects of sperm quality (particularly sperm viability and morphology) were less consistently reduced under nutrient limitation. Ejaculate traits tended to respond in a condition‐dependent manner to a wide range of dietary manipulations, especially to caloric and protein restriction. Finally, while all major taxa for which sufficient data exist (i.e. arthropods, mammals, fish) showed condition dependence of ejaculate traits, we detected some taxonomic differences in the life stage that is most sensitive to nutrient limitation, and in the degree of condition dependence of specific ejaculate traits. Together, these biologically relevant factors accounted for nearly 20% of the total variance in ejaculate responses to nutrient limitation. Interestingly, body size showed considerably stronger condition‐dependent responses compared to ejaculate traits, suggesting that ejaculate trait expression may be strongly canalised to protect important reproductive functions, or that the cost of producing an ejaculate is relatively low. Taken together, our findings show that condition‐dependence of ejaculate traits is taxonomically widespread, but there are also many interesting, biologically relevant sources of variation that require further investigation. In particular, further research is needed to understand the differences in selective pressures that result in differential patterns of ejaculate condition dependence across taxa and ejaculate traits.