OBJECTIVES: A recent randomized trial compared granulocyte colony-stimulating factor (G-CSF; filgrastim) to no G-CSF in elderly patients with aggressive NHL receiving CHOP chemotherapy [Osby, 2003]. A cost-effectiveness analysis is presented comparing CHOP alone to CHOP + G-CSF. METHODS: An economic model based on this trial compares the risk of neutropenia, disease relapse, and 5-year survival among patients receiving CHOP with or without G-CSF. Cost estimates were derived from published literature and data from U.S. health centers. Incremental cost-effectiveness ratios (ICERs) of $US/life year saved (LYS) were estimated, and sensitivity analyses performed. RESULTS: CHOP+G-CSF was associated with significantly fewer episodes of severe neutropenia (P<.001), FN (P<.001), greater dose intensity (P<0.05), fewer deaths (P=0.04), and improved 5-year survival (P=0.04). Based on five years of followup, the life years averaged 2.93 years in the CHOP alone group compared to 3.52 years in the CHOP+G-CSF arm. Expected costs were $41,400 and $39,747 for the G-CSF and control arms, respectively. Under baseline assumptions, the ICER for G-CSF support was estimated at $2769/LYS. Sensitivity analyses revealed G-CSF support to be cost saving across most plausible values for baseline FN risk, relative risk reductions for FN, infection-related mortality, and risks of disease relapse. G-CSF support remained cost saving until the control risk for disease relapse fell to <2%. Net cost savings were observed for FN relative risk reductions >56%. CONCLUSIONS: A recent clinical trial of G-CSF support demonstrated a reduction in neutropenic complications and improved survival. Incorporation of cost data into an economic model based on this trial demonstrates that G-CSF support is within accepted limits for cost-effectiveness across a broad range of assumptions.