One hundred and twenty-eight patients with non-Hodgkin's lymphoma (NHL), Hodgkin's disease (HD), and acute lymphoblastic leukemia (ALL) previously reported from a phase III trial of rhGM-CSF or placebo following autologous bone marrow transplantation (ABMT) were investigated for the development of late toxicities. Median follow-up is 36 months. No apparent long-term deleterious effects on BM function were observed. Moreover, disease-free survival and overall survival were similar for patients on both treatment arms, arguing for the long- term safety of recombinant human granulocyte macrophage-colony- stimulating factor (rhGM-CSF). The only factors predictive for both a high risk of relapse over time and mortality were having the diagnosis of ALL and/or undergoing ABMT in resistant relapse. We attempted to identify clinical variables before BM harvest, at the time of marrow infusion, or events within the first 100 days posttransplant, which might predict speed of neutrophil recovery in the setting of placebo or rhGM-CSF administration after ABMT. Only previous exposure to agents that deplete stem cells led to a significant delay in neutrophil recovery, suggesting their avoidance in patients who may undergo ABMT. Nevertheless, even those patients benefited from rhGM-CSF. For all patients, rhGM-CSF and agents that deplete stem cells were the strongest independent predictors for neutrophil engraftment. With the increasing use of newer hematopoietic growth factors both alone and in combination, long-term follow-up is essential to confirm the same safety that we report with rhGM-CSF.