Antigen receptor gene rearrangement studies are a sensitive means of determining lineage and clonality in lymphoproliferative disorders (LPDs) which remain difficult to classify after assessment of morphology and immunohistochemistry (IHC). This study investigates the utility of genotyping LPDs in a surgical pathology laboratory servicing a large teaching hospital. Ninety-eight specimens with detailed frozen (FS) and/or paraffin section IHC were studied, including 65 B-cell lymphomas, 14 T-cell lymphomas, 2 biopsies of T-zone dysplasia, one unclassifiable lymphoma, 8 Hodgkin's disease (HD) and 8 reactive nodes. Southern blotting (SE) was performed on tumor and control DNA cleaved with restriction enzymes EcoR1, Hind III and BamH1, using radiolabelled probes for the immunoglobulin heavy chain joining region, constant regions of kappa and lambda light chains, and the constant region of the T-cell receptor beta chain. All reactive nodes and those harbouring HD had DNA in the germline configuration, apart from J(H) rearrangement in one case each of HD and florid reactive hyperplasia. Of the non-Hodgkin's lymphomas (NHL), 17% did not reveal clonal rearrangements (11% B-NHL; 44% T-NHL). Most of the negative results could be explained by sampling error in partially involved nodes, highly polymorphous infiltrates where the neoplastic population may have been below the 1% threshold detectable by SE, and instances of anaplastic large cell lymphoma. After accounting for these cases, a 5% negative rate of genoclonality remained (3% B-NHL; 13% T-NHL). In the majority of NHL (95%), the diagnosis could be established on the basis of morphology and/or IHC alone. In only 4 cases was recourse to DNA analysis considered essential to distinguish between NHL and a reactive proliferation. In only 3 cases of NHL was genotyping needed to establish lineage (resolved in 2). In our laboratory SB is 5 times more expensive than FS-IHC (panel of 10 antibodies); we therefore suggest that SE should be restricted to those few cases where a diagnosis cannot be established by other means. It would be most cost-effective if such DNA studies were centralized in large laboratories.