We examined the three Alzheimer's disease (AD) family data sets for heterogeneity. Four markers that were represented in all three data sets were selected for analysis. Markers BCL3/EcoR-Mlu and D19S13/TaqI were chosen for chromosome 19 and D21S13/TaqI and D21S16/XbaI for chromosome 21. Homogeneity testing was performed on the data by use of Morton's pre-divided samples test (PS-test) and Smith's admixture test (A-test). The C-test of MacLean et al. [1992] was also used to test for linkage in the presence of heterogeneity. Assuming homogeneity, there was significant evidence of linkage to AD for BCL3, D19S13, D21S16 in the Duke data set and D19S13, D21S16 in the Boston data set. C-tests also provided evidence of linkage for BCL3 and D21S13 in the Duke data set, and D21S16 in the Boston data set. No evidence for heterogeneity within the data sets was found for any of the four markers using either the A-test or the C-test. For marker BCL3, PS-tests found evidence of heterogeneity between the three data sets and between early-versus late-onset families combined over all data sets.