Blacker et al.1 reported an association between Alzheimer disease (AD) and the deletion allele (A2M*2) of an intronic polymorphism in the -2-macroglobulin gene (A2M; 2), which confers a risk for AD (OR=3.55, 95% CI=1.90-7.03) comparable with that of APOE*E4 (OR=3.54, 95% CI=1.76-7.12). We analysed two independent sets of AD families3, 4 using the same family-based association (sibship disequilibrium test1 (SDT) and sib transmission-disequilibrium test5 (S-TDT)) methods. Following the scheme of Blacker et al.1, we limited these analyses to nuclear families of European descent in which all affected individuals had AD onset over 50 years, marker information was available for at least one unaffected sib and, in the case of the S-TDT, two or more distinct genotypes were present in the sibship. We averaged P values for the S-TDT over 100 iterations (10,000 replicates per iteration) using a Monte-Carlo method6. Both data sets demonstrated the association of AD and APOE*E4 (SDT, Duke, P=0.000007; Toronto, P=0.0009), indicating sufficient power to detect associations of the magnitude reported for A2M*2 (1). Although prior evidence shows that these pedigrees are enriched for an AD locus on chromosome 12 near A2M (Refs 3,7), we were unable to detect an association with the A2M*2 polymorphism using either the SDT or S-TDT, even when the "stringent unaffecteds" method1 was applied (SDT, Duke, n = 60, P = 0.80, Toronto, n = 45, P = 0.82; S-TDT, Duke, n = 17, P = 0.64, Toronto, n = 21, P = 0.75). Furthermore, we did not detect an association in two independent series of sporadic AD cases of European descent after adjustment for APOE*E4 status (Table 1), despite the fact that each data set had more than 80% power to detect an odds ratio as low as 1.87 (an effect much smaller than reported for A2M*2; 1).