Polymerase chain reaction (PCR) amplification of the t(14;18) has been shown to be a highly sensitive method to detect minimal residual disease in patients with non-Hodgkin's lymphoma (NHL) whose tumors bear this translocation. The ideal tissue source to detect residual lymphoma would be from a previously involved lymph node. However, lymphoid tissue is rarely available once patients achieve complete remission. Although PCR amplification has been used to detect residual lymphoma cells in both bone marrow (BM) and peripheral blood (PB) of patients in complete remission, it is presently unknown whether BM and PB are equivalent tissue sources to detect residual disease. In the present study, we compared the clinical utility of the detection of residual lymphoma in both the BM and the PB of patients with advanced-stage non-Hodgkin's lymphoma before, at the time of, and after high-dose therapy and autologous BM transplantation (ABMT). The detection of residual lymphoma in either the BM or PB was associated with decreased disease-free survival. However, in the present study, 44% of patients who relapsed had no evidence of circulating lymphoma cells in their PB. At the time of BM harvest, PCR-detectable residual lymphoma cells were detected in 211 of 212 patients; although, in a subset of these patients analyzed, lymphoma cells were detected in the peripheral blood of only 49% of patients. When residual lymphoma cells within the autologous BM are infused into the patient these cells are rapidly detectable circulating in the PB in the patient. These cells continue to circulate during the immediate posttransplant period and be detectable in the PB in the majority of patients who are infused with marrow containing residual lymphoma. We conclude that BM is a more informative tissue source than PB in detecting minimal residual disease at the time of and after ABMT, and that contamination of PB early after ABMT appears to be the consequence of reinfusion of lymphoma cells within autologous marrow.