Crossbreeding parameters of immune response traits were estimated from a set of well characterized crossbred populations derived from three chicken lines selected over 12 generations for three different general immune response traits and their F1, F2 and backcrosses. The three traits investigated were the selection criteria from each of the lines, i.e. antibody response to the Newcastle disease virus vaccine 3 weeks after vaccination (ND3), cell-mediated immune response (response to phytohemagglutinin, PHA) and phagocytic activity measured as carbon clearance (CC). Crossbreeding parameters included direct and maternal additive line effects, direct and maternal heterosis as well as direct epistatic recombination loss. They were estimated as linear combinations of genetic group effects estimated using animal model methodology. Significant line differences were obtained for ND3 and, to a lesser extent, CC. They were mainly due to direct effects, maternal effects being significant for none of the 3 traits. Significantly negative direct heterosis effects were also observed for ND3 and CC, but not for PHA. Maternal heterosis effects were not estimated for CC. They were non significant for PHA, and negative and significant (− 0.78 ± 0.24) for ND3. The significant favourable recombination gain estimated for ND3 (3.21 ± 0.88) indicates that epistatic interactions could be important for this trait. The present work shows that it was worthwhile to complete second generation crosses to be able to assess to what extent immunity gained by selection is maintained in advanced crossbred generations, and to compare the transmission of immune traits implicated in different aspects of immunity.