The diagnosis of gastrointestinal (GI) lymphoid infiltrates can be challenging when based only on conventional microscopic assessment. When marked cytologic atypia is present, a diagnosis of malignant neoplasm is readily made; however, the distinction between a low-grade malignant neoplasm and a reactive process is much more difficult. If unfixed tissue is available, immunohistologic or genotypic methods that are usually aimed at defining B-lymphocytic monotypism can be applied. However, paraffin-embedded tissue has generally been deemed unsuitable for these techniques. We assessed the value of a panel of immunohistochemical stains and a seminested polymerase chain reaction (PCR) for the analysis of lymphoid infiltrates in routinely processed GI biopsy specimens from 49 archival cases, including morphologically benign, indeterminate, and overtly malignant lesions. Clinical outcome was used as the retrospective diagnostic standard; end points were death (of lymphomatous disease or otherwise) and clinical evidence of lymphoma. According to light microscopic criteria, 19 cases were classified as benign, 17 as malignant, and 13 as atypical. Immunophenotyping correctly identified 28 of 31 benign and 14 of 18 malignant lesions (7 cases had an indeterminate immunoprofile). Genotypic analysis correctly identified 12 of 18 malignant and 29 of 31 benign lesions, but spurious monoclonal bands were produced by PCR amplification of 2 of the latter 31 cases. No single technique exists for correct categorization of all paraffin-embedded specimens of GI lymphoid infiltrates. We recommend a sequential approach to the use of available diagnostic modalities.