The use of peripheral blood rather than marrow has potential advantages for monitoring minimal residual disease during the treatment of leukaemia. To determine the feasibility of using blood, we used a sensitive polymerase chain reaction method to quantify leukaemia in the blood and marrow in 35 paired samples from 15 children during induction treatment. Leukaemic cells in the blood ranged from 1.1 x 10(-2) to < 9.4 x 10(-7) leukaemic cells/total cells, corresponding to 1.3 x 10(7) to < 2 x 10(3) leukaemic cells/l. In 15 paired samples, leukaemia could be quantified in both tissues and in 20 paired samples, leukaemia was not detected in one or both tissues so that only upper level limits could be set. In the former 15 pairs, the level of leukaemia in peripheral blood was directly proportional to that in marrow but was a mean of 11.7-fold lower. Leukaemia in blood was detected in 10/12 pairs in which the level in marrow was > 10(-4), but in only two of 13 pairs in which the level in marrow was < 10(-5). Patients studied at multiple time-points showed parallel declines in the number of leukaemic cells in both tissues. The results showed that leukaemia could be monitored in peripheral blood during induction therapy, and quantitative considerations based on the results suggest that monitoring of blood during post-induction therapy may be of value in detecting molecular relapse.