The utility of immunohistochemistry (IHC) as a screening method for the identification of persons with mutations in the DNA mismatch repair (MMR) genes in hereditary nonpolyposis colorectal cancer (HNPCC) remains to be defined. In this study, we analyzed the value of IHC versus that of microsatellite instability (MSI) testing in predicting mutation status of the MLH1, MSH2, and MSH6 genes in colorectal carcinomas and adenomas, and explored the frequency and significance of immunohistochemical staining variability. The study samples included 83 carcinomas and 29 adenomas derived from 110 patients who had strong family histories of colorectal cancer. Our results showed that IHC correctly predicted MSI status in 76% of the cases with a specificity of 100%. The overall sensitivity of IHC in predicting a germline mutation was 79% (30 of 38) with a specificity of 89% (48 of 54), whereas that of MSI testing was 97% (30 of 31) with a specificity of 83% (35 of 42). Six of 31 analyzable cases that had a disease-causing mutation and exhibited MSI showed normal IHC. The lower sensitivity of IHC was caused mainly by its low sensitivity in detecting MLH1 gene mutation (4 of 9). Coexisting adenomas and carcinomas observed in the same slide (n=12) showed a similar or identical staining pattern for all three proteins. No significant difference was detected in the sensitivity of IHC or MSI in detecting a germline mutation between isolated adenomas and carcinomas. In IHC-positive cases, heterogeneous staining was noted in 30% to 40% of the cases with the three different antibodies, and cytoplasmic staining in 5% to 13%. Weak IHC (defined as positive staining in <10% of the tumor with weak intensity) was noted in 14 tumors: 5 for the MLH1 antibody, 1 for MSH2, and 8 for MSH6. One of the 5 MLH1 cases exhibited MSI and had an MLH1 germline mutation. Five of the 8 MSH6 cases exhibited MSI and had MSH2 germline mutations. In conclusion, our study shows that 1) IHC identifies a significant portion of colorectal tumors derived from MMR gene germline mutation carriers and can be used as an adjunct measure in the identification of HNPCC families, but IHC cannot replace MSI testing; 2) adenomas have similar MMR protein expression patterns as carcinomas and may serve as an adequate sample for screening purposes in the identification of patients with MMR mutations; 3) not all IHC-positive cases show uniform positivity throughout the tumor; and 4) weak and focal staining of an MMR protein may be associated with MSI or gene mutation or both, suggesting the need to incorporate staining intensity in further IHC studies.