Missense variants of DNA mismatch repair (MMR) genes pose a problem in clinical genetics as long as they cannot unambiguously be assigned as the cause of Lynch syndrome (LS). To study such variants of uncertain clinical significance, we have developed a functional assay based on direct measurement of MMR activity in mouse embryonic stem cells expressing mutant protein from the endogenous alleles. We have applied this protocol to a specific truncation mutant of MSH2 that removes 60 C-terminal amino acids and has been found in suspected LS families. We show that the stability of the MSH2/MSH6 heterodimer is severely perturbed, causing attenuated MMR in in vitro assays and cancer predisposition in mice. This mutation can therefore unambiguously be considered as deleterious and causative for LS.