Increasing numbers of transplants, particularly autotransplants, are performed using blood-derived cells. Most subjects have leukemia but others have solid tumors or even genetic disorders. Despite this expanding data base, several important issues are unresolved. Can blood-derived cells reconstitute short-term hematopoiesis? Here, the answer is likely yes, although this is not yet proven. Studies using genetically marked cells should resolve this issue. Next, can blood-derived cells reconstitute long-term hematopoiesis? Here, the answer is unknown, but it would not be surprising were this not so. However, since most (if not all) high-dose chemotherapy and radiation treatments do not completely eradicate endogenous hematopoietic stem cells, reconstitution of long-term hematopoiesis from the graft may not be necessary for blood-derived grafts to be useful clinically. Another unresolved area is whether blood-derived grafts have a lower likelihood of tumor recurrence because of a lower probability of tumor contamination (qualitative or quantitative) or because of a different cellular composition of the graft. This issue is only answerable in controlled trials. Most data suggest that recurrence rate would not be higher than bone marrow-derived grafts. Whether it is lower is unknown. Analysis of this point is confounded by the fact that most (perhaps all) relapses that occur post-transplant using current conditioning schedules are explicable by residual leukemia in the recipient. Thus, tumor contamination of the graft is not presently an operationally important issue. There are some recent developments in this area. One is the use of umbilical cord blood cells to reconstitute hematopoiesis in a child with Fanconi anemia. Additional data are needed to evaluate this approach.(ABSTRACT TRUNCATED AT 250 WORDS)