AbstractSelective breeding for the character “agglutinin production to heterologous erythrocytes” was performed on random‐bred populations of albino mice. Two types of selection were carried out: Selection I: the first 6 generations were immunized with sheep erythrocytes (SE), then SE and pigeon erythrocytes (PE) were alternated at each generation to avoid the interference of maternal antibody. Selection II: all generations were immunized with SE 60–70 days after weaning in order to eliminate the maternal antibody. Both were selected for the character agglutinin titer 14 days after immunization using an optimal dose of erythrocytes.Selection I.The mean response to SE of the foundation population was 7.36 ± 1.60. The selection limit was attained between the 15th and the 20th generation. The F20−F30were considered as homozygous for the character investigated. The mean SE agglutinin titers were 9.15 (1/2820) in the high responder line and 2 (1/20) in the low responder line. This corresponds to 140‐fold interline difference in agglutinin titers. SE agglutinins determined in F1hybrids, F2hybrids and backcrosses obtained from F23− F29demonstrated incomplete dominance of high responsiveness (23–29%).The inbreeding coefficient produced by close colony breeding was 0.55 in high and 0.71 in low responders at the 20th generation when both lines were homozygous with respect to agglutinin synthesis.Selection II.The mean response of the foundation population to SE was 7.79 ± 1.56. The selection limit was attained in F14−F17generations that were considered homozygous for the character investigated. The mean SE agglutinin titer was 9.2 (1/2900) in high responders and 2.8 (1/35) in low responders, which is an 83‐fold interline difference in agglutinin titers. The results of both selections indicate that the character agglutinin synthesis is subject to polygenic regulation. The heritability (h2) of this character was estimated to be between 0.18 and 0.36 (range between the extreme values of Selections I and II).The attempt to evaluate approximately the number of “loci” is compatible with the hypothesis that a group of about 10 loci regulates the quantitative antibody response. The sources of errors of this evaluation and the significance of the