A clonal-specific polymerase chain reaction technique to detect T-cell receptor delta gene rearrangement in acute lymphoblastic leukaemia (ALL) and non-Hodgkin's lymphoma (NHL) was evaluated. It was applied to detect minimal residual disease. A sensitive and specific technique to detect minimal residual disease for T-cell malignancies was explored. Southern analysis and polymerase chain reaction (PCR) were used to detect the rearranged V-D-J segment of T-cell receptor delta (TCRG) gene from malignant cell specimens of patients with leukemia and lymphoma of T-cell lineage. The PCR product was sequenced and from the DNA sequences of the V-D-J region, a 3' anti-sense primer was designed and synthesized for clonal specific PCR (CS-PCR). T-cell receptor delta (TCR-6) gene rearrangement was studied in 40 cases of acute leukaemia and lymphoma of Tcell lineage at diagnosis. Using Southern analysis, the positive rates were 28 and 32% for the 18 T-lymphoma and 22 T-ALL, respectively. A one stage Polymerase Chain Reaction (PCR) technique was used to detect the rearrangement in Southern positive cases and the PCR positive rates were 80 and 86%, respectively. The PCR technique had a sensitivity of 0.1%. Serial follow-up marrow specimens were available from 4 T-ALL patients following chemotherapy for monitoring of minimal residual disease. Their PCR products were DNA sequenced. A 3' primer was designed for each case for a clonal specific (CS) PCR. The technique had a sensitivity of 0.003%. It was applied to detect minimal residual disease in serial follow-up marrow samples. The first patient had persistent negative CS-PCR results and enjoyed continuous remission for more than 3 years. The second patient with negative one stage PCR but positive CS-PCR results had eventual relapse of leukaemia. The other two patients never achieved a morphological remission. These preliminary results appeared to support the usefulness of these PCR techniques in detecting minimal residual disease and predicting relapses for ALL. However, further clinical correlation in larger populations of patients is necessary.