We prospectively evaluated the effects of celecoxib (CEL) on the gastrointestinal (GI) tract of rheumatoid arthritis (RA) patients with endoscopically identified GI mucosal injury after therapeutic switching from the long-term use of traditional nonsteroidal anti-inflammatory drugs (NSAIDs). Upper GI endoscopy was performed on RA patients who had been treated with NSAIDs for ≥3 months. GI mucosal injury was evaluated according to the modified LANZA score. Patients with mucosal injury without ulcers were switched from NSAIDs to CEL, while those with ulcers were switched to CEL with famotidine after ulcer healing. At week 16 of treatment, GI mucosal injury was endoscopically revaluated. An efficacy analysis was performed before therapeutic switching and at 8 and 16 weeks post-switching. Endoscopic analysis revealed GI mucosal injury, including six ulcers, in 45 of the 82 patients (54.9%). Sixteen weeks after switching to CEL, LANZA scores were significantly improved [2.1 ± 0.8 (pre-switching) vs. 1.6 ± 1.3, P = 0.0073] in patients with LANZA scores of 1, 2, or 3 (n = 35). The Disease Activity Score using 28 joint counts (DAS28) [erythrocyte sedimentation rate item score (ESR4) (P = 0.0257) and C-reactive protein item score (CRP4) (P = 0.0031)] was also significantly improved by week 16. Based on these results, we conclude that preexisting NSAID-induced upper GI injury is improved following therapeutic switching to CEL without any reduction in analgesic efficacy.