The pathogenesis of patient dissatisfaction following involuntary therapeutic substitutions involving proton pump inhibitors (PPIs) is poorly understood. The aim of this study was to describe the patient population experiencing therapeutic failure and investigate whether failure was related to individual differences in response to the different PPIs. Treatment failures in a lansoprazole-rabeprazole therapeutic substitution program were compared to switch successes. A subgroup was randomized in a double-blind, double-dummy, crossover study to four 2-week periods of lansoprazole-rabeprazole-lansoprazole-rabeprazole or vice versa. Measures included overall rating of gastrointestinal reflux disease (GERD) symptoms for the past week as well as the frequency and distress scales of the GERD Symptom Assessment Scale. One hundred fifteen nonresponders were compared with 54 successful responders. Nonresponders consisted primarily of patients with GERD (74%, vs. 44% of responders; P = 0.005) who had undergone upper gastrointestinal endoscopy (50%, vs. 31% of responders; P = 0.02). Twelve patients completed the randomized treatment study. The interrater kappa coefficient for responder status was estimated to be 0.80 for lansoprazole and 0.21 for rabeprazole. The majority of PPI nonresponders had a clinical diagnosis of GERD and were receiving >/=40 mg of rabeprazole daily. This pilot study provides new insights into the design of subsequent studies of nonresponders in PPI therapeutic substitution programs.