OBJECTIVES: Budget Impact Analysis (BIA) preformed for new and expensive anti-cancer drugs should rely on precise data regarding the cost and amount of drug that would be utilized. Besides epidemiology & required doses, BIA needs to take into account the mean duration of treatment (DOT). In metastatic cancer, patients are usually treated by each line until disease progression. When reporting phase III trials, not all patients have failed treatment, and only "median time to treatment failure" is available. Therefore, BIA would rely on the median rather than the mean DOT. Five years after introducing Trastuzumab in the Israeli national formulary, we evaluated whether the original BIA based on a phase III trial, was accurate. METHODS: Clalit Health Services' computerized database was used to determine Trastuzumab DOT in all patients that started treatment during 2000-2002. The mean & median DOT was compared to the "median time to treatment failure" (5.8 months) in a phase III trial. RESULTS: In year 2000, the median and average DOT were 6.0 months and 9.9 months, respectively (n=132) (patients have received prior treatment with AC & T, as determined by formulary criteria). In 2001 the median and average DOT were 9.0 and 13.6 months, respectively (n=89) (criteria were changed, and patients were only required to be pretreated with AC). In 2002, the median DOT was 11.0 months and the mean DOT was14.4 months (n=71). CONCLUSIONS: In "real life" patients, mean trastuzumab DOT for metastatic breast cancer was 31-65% higher than the median DOT in the same patients. Median DOT itself was up to 55% higher than reported in the phase III trial. When performing BIA for treatments of metastatic cancer, the estimation of the budget impact may be substantially higher when using the mean rather the median value, reported in phase III clinical trials.