Transactional (‘optimal skew’or concessions') models of social evolution emphasize that dominant members of society can be favoured for donating parcels of reproduction to same–sexed subordinates in return for cooperation by the latter. We developed a mathematically similar model in which extra–pair paternity in broods receiving biparental care is viewed as emerging from a reproductive transaction between the paired mates. The model quantitatively predicted the maximum paternity that a male mate can demand before its female mate is favoured to break the pair bond and caring solitarily for a brood sired entirely by a neighbouring male. The model predicts that extra–pair paternity results when the neighbouring male is of sufficiently higher quality than the male mate. In such cases, the exact amount of extra–pair paternity will vary directly with the difference in quality between the two males and inversely with the value (fitness impact) of the male mate's parental care. Importantly, the transactional model provided a unified explanation for experimental and observational evidence that extra–pair paternity rises with decreasing quality of the male mate, increasing genetic variability among breeding males, increasing breeding density, increasing availability of food and decreasing involvement of the male mate in parental care.