This study addresses the utility of 11 blood groups as selection aids in Holstein breeding schemes and considers issues inherent to the approach of resolving quantitative variation into components that are due to quantitative trait loci. The data consisted of predicted transmitting abilities of 22,614 bulls, first lactation information on 1,924,171 cows, and type scores on 447,800 cows. Linear models were fitted under male half-sib designs, female half-sib designs, and granddaughter designs as well as under the assumption of direct effects of the markers. The evolution of allele frequencies through time was determined, and previous research results were synthesized according to criteria of consistency of biological significance. The inconsistency of results across studies and analytical designs alludes to the importance of the intrinsic nonadditivity of genetic and biological phenomena to quantitative trait locus detection and marker-assisted selection. In our analyses, three associations met the criteria of consistency--a C blood group effect on rump angle, an L effect on milk yield and composition traits, and an S effect on milk fat yield. The M locus appears to be directly associated with effects on milk and protein yields. An enhanced understanding of the biochemical and physiological bases of quantitative genetics should be a long-term objective of this type of genetic analysis.