A matrix decomposition model for analyzing prehistoric intracemetery biological variability is presented. The model, based on the correlation between interindividual biological and burial distance matrices, provides an estimate of the number of distinct burial populations interred within a single cemetery, which effectively ameliorates identifiability problems typical of general variability analyses (Raemsch  Ph.D. dissertation, SUNY at Albany). Unlike other methods for analyzing intrasite variability analysis (e.g., kinship analysis and microchronology), this method is not reliant on a priori subgroup definition. Assuming a kin-structured burial pattern, a residual matrix is defined based on the raw data matrices; the proportion of negative or zero residuals to the total number of residuals provides an estimate of the number of hidden subgroupings (lineages) within the cemetery. The comparative utility of the model is demonstrated on a series of protohistoric and historic period skeletal samples from the Georgia coast, where it demonstrated that the increase in phenotypic variability during the early mission period is the result of population aggregation at the missions and not necessarily genetic admixture between local and nonlocal groups. Am J Phys Anthropol 121:000–000, 2003. © 2003 Wiley-Liss, Inc.