The important question of whether residual tumor in the bone marrow or peripheral blood stem cell graft contributes to relapse in autologous bone marrow transplantation in patients with non-Hodgkin's lymphoma can be addressed only if there is an accurate and sensitive measurement of tumor cell contamination of the graft. Assays utilizing DNA amplification based on the polymerase chain reaction (PCR) are highly sensitive. Tumor-specific primers and probes can be designed for the clonally rearranged Ig or T cell antigen receptor genes in the original tumors, and these can then be used to detect minimal residual disease in subsequent specimens. Specific translocations can also be exploited as tumor markers, and the t(14;18) translocation has been widely employed for detecting tumor cells in blood and bone marrow samples. Lymphoma cells have also been grown successfully in tissue culture, and the detection of tumor contamination of autologous grafts has been associated with a poorer prognosis in patients with intermediate- or high-grade lymphoma. It is of interest to compare the sensitivity of tumor detection and the predictive value for patient survival of the PCR-based and culture-based assays. The information obtained may help to determine whether minimal tumor contamination of an autologous graft is clinically significant and, if so, the assay(s) that should be employed.