Universal screening for Lynch syndrome in colorectal cancer is recommended, and immunohistochemistry for the mismatch repair proteins is commonly used. To reduce cost, some screen using only MSH6 and PMS2, with reflex to the partner stain if either are absent (two-stain method). An expression pattern revealing absent MSH2 and intact MSH6 is not expected, but could result in failed Lynch syndrome detection. We analyzed tumors with absent MSH2 but any degree of MSH6 expression to determine if the two-stain method could miss MSH2 mutations. One-thousand seven-hundred thirty colorectal cancer patients from the Ohio Colorectal Cancer Prevention Initiative underwent tumor screening using microsatellite instability and immunohistochemistry. The two-stain method was used for 1235 cases; staining for all four proteins was completed for 495 cases. The proportion of positive cells and staining intensity were reviewed for MSH6, as well as MSH2 when available. Patients with mismatch repair deficiency underwent next-generation sequencing of germline DNA for mismatch repair genes. If negative, tumor next-generation sequencing was performed to assess for somatic mutations. Overall, thirty-three (1.9%, 33/1730) MSH2-absent cases were identified. Of those, fourteen had no MSH6 expression but eight (0.5%, 8/1730) had ambiguous and eleven (0.6%, 11/1730) had convincing MSH6 expression that could have been interpreted as intact. Germline next-generation sequencing identified MSH2 mutations in 11/14 cases with absence of both stains, 7/8 cases with ambiguous MSH6 expression, and 9/11 cases with convincing MSH6 expression. All remaining cases, except one, had double somatic mutations. The two-stain method fails to detect some patients with Lynch syndrome: (1) significant staining weaker than the control may be incorrectly interpreted as intact MSH6, or (2) Weak or focal/patchy MSH6 can be retained with the absence of MSH2. Accordingly, we recommend the four-stain method be used for optimal Lynch syndrome screening detection.