Resistance to pathogens is often invoked as an indirect benefit of female choice, but experimental evidence for links between father's sexual success and offspring resistance is scarce and equivocal. Two proposed mechanisms might generate such links. Under the first, heritable resistance to diverse pathogens depends on general immunocompetence; owing to shared condition dependence, male sexual traits indicate immunocompetence independently of the male's pathogen exposure. By contrast, other hypotheses (e.g. Hamilton-Zuk) assume that sexual traits only reveal heritable resistance if the males have been exposed to the pathogen. The distinction between the two mechanisms has been neglected by experimental studies. We show that Drosophila melanogaster males that are successful in mating contests (one female with two males) sire sons that are substantially more resistant to the intestinal pathogen Pseudomonas entomophila-but only if the males have themselves been exposed to the pathogen before the mating contest. By contrast, sons of males sexually successful in the absence of pathogen exposure are less resistant than sons of unsuccessful males. We detected no differences in daughters' resistance. Thus, while sexual selection may have considerable consequences for offspring resistance, these consequences may be sex-specific. Furthermore, contrary to the 'general immunocompetence' hypothesis, these consequences can be positive or negative depending on the epidemiological context under which sexual selection operates.