The objective of the study was to estimate the net impact on health resource utilisation of using recombinant granulocyte colony-stimulating factor (filgrastim) following myelosuppressive chemotherapy. Cost minimisation of the study medication in a randomised, double-blind, placebo-controlled clinical trial was conducted in teaching institutions and affiliated community hospitals participating in a clinical trial. 68 patients with small cell lung cancer undergoing cyclophosphamide, doxorubicin and etoposide chemotherapy were randomised to blinded placebo or filgrastim study medication at three or 14 clinical trials sites. The patients received daily subcutaneous injections of filgrastim or placebo, initiated 24 h after chemotherapy and continued until the neutrophil count exceeded 10,000 x 10(6)/l after the time of the expected nadir. Differences in total charges, costs and Medicare payments between treatment groups were the main outcomes measured. Compared to placebo patients, filgrastim-treated patients had significantly fewer and less resource-intensive hospitalisations. After accounting for filgrastim purchase and administration, the charge model predicts overall savings from filgrastim use in a clinical setting in which the risk of febrile neutropenia is high for patients not receiving filgrastim. The Medicare and cost models predict only a partial recapture of the cost of filgrastim therapy. The health care resources impact of filgrastim was sensitive to the risk of hospitalisation with febrile neutropenia, and to the perspective chosen for measuring resource utilisation (charges, costs or Medicare payments). The adjunctive use of filgrastim following myelosuppressive chemotherapy leads to partial or complete recapture of the cost of purchasing and administering the product.