Background and Aims: The best initial treatment of human immunodeficiency virus (HIV)-infected patients with esophageal symptoms is unknown. The outcome, including safety and cost-effectiveness, of fluconazole compared with endoscopy as a treatment strategy for HIV-infected patients with new- onset esophageal symptoms was evaluated. Methods: During a 53-month period, 134 HIV-infected patients with esophageal symptoms were randomized prospectively to groups receiving either standard doses of fluconazole or endoscopy. Results: Among the 68 patients in the fluconazole group, a complete symptomatic response was observed in 56 patients (62%), usually within 1 week. The most common endoscopic findings in the 66 patients in the endoscopy group included Candida esophagitis alone in 42 patients (64%) and ulcerative esophagitis in 10 patients (15%). Patients responding to empirical antifungal therapy or who had Candida esophagitis alone at endoscopy were less likely to have severe symptoms (P = 0.027) or odynophagia as the only symptom (P < 0.001) but more frequently had odynophagia and dysphagia (P = 0.007) and thrush (P = 0.002). Empirical fluconazole was cost-effective, saving $738.16 per patient. Conclusions: Empirical oral antifungal therapy with fluconazole is highly efficacious, safe, and cost-effective for HIV- infected patients with new-onset esophageal symptoms.