Recombinant human granulocyte colony-stimulating factor (G-CSF; filgrastim) shortens the time to neutrophil recovery after intensive chemotherapy, but its role in the treatment of adults with acute lymphoblastic leukemia (ALL) is uncertain. We randomly assigned 198 adults with untreated ALL (median age, 35 years; range, 16 to 83) to receive either placebo or G-CSF (5 microgram/kg/d) subcutaneously, beginning 4 days after starting intensive remission induction chemotherapy and continuing until the neutrophil count was >/=1, 000/microL for 2 days. The study assignment was unblinded as individual patients achieved a complete remission (CR). Patients initially assigned to G-CSF then continued to receive G-CSF through 2 monthly courses of consolidation therapy. Patients assigned to placebo received no further study drug. The median time to recover neutrophils >/=1,000/microL during the remission induction course was 16 days (interquartile range [IQR], 15 to 18 days) for the patients assigned to receive G-CSF and 22 days (IQR, 19 to 29 days) for the patients assigned to placebo (P < .001). Patients in the G-CSF group had significantly shorter durations of neutropenia (<1, 000/microL) and thrombocytopenia (<50,000/microL) and fewer days in the hospital (median, 22 days v 28 days; P = .02) compared with patients receiving placebo. The patients assigned to receive G-CSF had a higher CR rate and fewer deaths during remission induction than did those receiving placebo (P = .04 by the chi-square test for trend). During Courses IIA and IIB of consolidation treatment, patients in the G-CSF group had significantly more rapid recovery of neutrophils >/=1,000/microL than did the control group by approximately 6 to 9 days. However, the patients in the G-CSF group did not complete the planned first 3 months of chemotherapy any more rapidly than did the patients in the placebo group. Overall toxicity was not lessened by the use of G-CSF. After a median follow-up of 4. 7 years, there were no significant differences in either the disease-free survival (P = .53) or the overall survival (P = .25) for the patients assigned to G-CSF (medians, 2.3 years and 2.4 years, respectively) compared with those assigned to placebo (medians, 1.7 and 1.8 years, respectively). Adults who received intensive chemotherapy for ALL benefited from G-CSF treatment, but its use did not markedly affect the ultimate outcome.