The effects of a clinical protocol for filgrastim use in oncology patients were studied. A chart review was conducted for every fourth oncology inpatient who received filgrastim at a community hospital between January and June 1996 to determine how filgrastim was being used in the hospital's oncology patients. The results were presented to the oncology committee, and a filgrastim protocol was implemented. The protocol stated that filgrastim would be discontinued when the absolute neutrophil count (ANC) was > or =1500 cells/mm3 for two days after the neutrophil nadir. Six months after the protocol was implemented, a follow-up evaluation was conducted by reviewing the chart for every fourth oncology patient who received filgrastim between November 1996 and April 1997. Twenty-one patient charts were reviewed before the protocol was implemented, and 34 charts were reviewed after implementation. The results showed there was compliance with the protocol for 19 (76%) of the 25 patients evaluable for compliance. Sixty-seven percent of patients were febrile before the protocol was implemented, and 56% were febrile afterward. Ten percent of patients had documented infections before implementation, compared with 12% afterward. The average ANC at which filgrastim was discontinued before and after the protocol was implemented was 6839 and 5538 cells/mm3, respectively. Filgrastim was discontinued by a pharmacist in 32% of cases. A $22,416 cost saving was achieved in the first six months after protocol implementation, with no compromise in clinical efficacy. A pharmacy-based protocol for discontinuing filgrastim therapy in oncology patients saved a community hospital more than $22,000 in the first six months withno adverse impact on the drug's effectiveness.