We examined the validity of immunohistochemistry for mismatch repair (MMR) proteins in colorectal cancer specimens to identify patients at risk for Lynch syndrome (hereditary nonpolyposis colorectal cancer) and patients with sporadic microsatellite instable colorectal cancer. This was assessed by observer agreement for and accuracy of interpretation of immunohistochemistry. Seven pathologists from 5 different pathology laboratories evaluated 100 molecularly defined colorectal cancers stained for MLH1, PMS2, MSH2, and MSH6. Two of the pathologists were experienced in interpretation of immunohistochemistry for MMR proteins. After evaluation of a subset of 20 cases, a discussion meeting was organized, after which pathologists evaluated all 100 cases. Staining patterns were interpreted as aberrant, normal, or indefinite. In 82% of tumors, 5 or more pathologists reached the same interpretation, which was considered the consensus diagnosis. Consensus was reached slightly less frequently in microsatellite instable than in stable tumors, and interobserver variation was moderate to substantial (kappa: 0.49-0.79). More microsatellite instable tumors showed an indefinite staining pattern compared with microsatellite stable tumors. Three out of 7 pathologists, including the 2 experienced pathologists, did not miss a microsatellite instable tumor. Each pathologist found at least 1 tumor with an aberrant staining pattern, whereas consensus was a normal staining pattern and the tumor was microsatellite stable. We conclude that, if restricted to experienced pathologists, immunohistochemistry is a valid tool to identify patients at risk for Lynch syndrome and patients with sporadic microsatellite instable colorectal cancer. An indefinite or aberrant staining result has to be followed by molecular microsatellite instability analysis to confirm the presence of a defective DNA MMR system.