OBJECTIVE: Number needed to treat (NNT) provides an estimate of the number of patients who need to be treated with a drug to avoid an adverse outcome on alternate therapy. This quantitative analysis focused on treatment responses to proton pump inhibitors according to disease severity. METHODS: Efficacy data from four clinical trials using once-daily esomeprazole 40 mg compared with omeprazole 20 mg (n = 3) and lansoprazole 30 mg (n = 1) for treatment of erosive esophagitis (EE) were identified. EE was graded A-D using the Los Angeles (LA) classification. For each trial we calculated the therapeutic gain, the absolute risk reduction (ARR) and the number needed to treat (NNT) for healing EE at week 8, for all patients (LA Grades A-D) and separately for those with severe disease (LA Grades C & D). RESULTS: For all grades of esophagitis, the therapeutic gain achieved with esomeprazole was 9.5%, 7.2% and 2.4% versus omeprazole, and 3.8% versus lansoprazole. The NNT with esomeprazole for severe disease ranged between 5 and 10 relative to omeprazole. The NNT with esomeprazole for severe disease was 8 relative to lansoprazole, indicating that for every 8 patients treated with esomeprazole, 1 treatment failure with lansoprazole may be prevented. CONCLUSIONS: Treatment with esomeprazole provided therapeutic gain regardless of the baseline severity of EE compared with lansoprazole and omeprazole. Because the severity of clinical symptoms is not predictive of disease severity, treatment with the most effective agent appears to be a rational therapeutic decision as supported in this evidence-based approach.