Kelvin is a flexible program for measuring statistical evidence based on genetic data. It is based on the PPL framework and produces output on the posterior probability (0,..,1) scale. Primary outcome statistics include the PPL (posterior probability of linkage (L)), as well as various forms of the PPLD (posterior probability of linkage disequilibrium (LD)). In the context of analysis between a trait locus and a marker, the PPLD models allelic association due to very close genomic proximity. Kelvin handles both dichotomous and quantitative trait data, and can accept as input multigenerational pedigree data, nuclear families, “trios,” and case-control data. The program supports parent-of-origin effects, co-variate dependent penetrances (or means), and two-locus epistasis. Kelvin also includes special routines for sequentially updating results in order to accumulate evidence across multiple sets of data.