In order to evaluate the potential clinical and economic benefits of granulocyte colony-stimulating factor (G-CSF, filgrastim) following peripheral blood progenitor cells (PBPC) rescue after high-dose chemotherapy (HDCT), 23 consecutive patients aged less than 60 years with poor-prognosis, high-grade non-Hodgkin's lymphoma (NHL) were entered into a prospective randomized trial between May 1993 and September 1995. Patients were randomized to receive either PBPC alone (n = 12) or PBPC+G-CSF (n = 11) after HDCT with busulphan and cyclophosphamide. G-CSF (300 microg day[-1]) was given from day +5 until recovery of granulocyte count to greater than 1.0 x 10(9) l(-1) for 2 consecutive days. The mean time to achieve a granulocyte count > 0.5 x 10(9) l(-1) was significantly shorter in the G-CSF arm (9.7 vs 13.2 days; P<0.0001) as was the median duration of hospital stay (12 vs 15 days; P = 0.001). In addition the recovery periods (range 9-12 vs 11-17 days to achieve a count of 1.0 x 10(9) l[-1]) and hospital stays (range 11-14 vs 13-22 days) were significantly less variable in patients receiving G-CSF in whom the values clustered around the median. There were no statistically significant differences between the study arms in terms of days of fever, documented episodes of bacteraemia, antimicrobial drug usage and platelet/red cell transfusion requirements. Taking into account the costs of total occupied-bed days, drugs, growth factor usage and haematological support, the mean expenditure per inpatient stay was pound sterling 6500 (range pound sterling 5465-pound sterling 8101) in the G-CSF group compared with pound sterling 8316 (range pound sterling 5953-pound sterling 15,801) in the group not receiving G-CSF, with an observed mean saving of 1816 per patient (or 22% of the total cost) in the G-CSF group. This study suggests that after HDCT and PBPC rescue, the use of G-CSF leads to more rapid haematological recovery periods and is associated with a more predictable and shorter hospital stay. Furthermore, and despite the additional costs for G-CSF, these clinical benefits are not translated into increased health care expenditure.