HIV-1-infected patients frequently have opportunistic esophageal infections which, when associated with severe immunodeficiency, can be attributed to unusual pathogens. The clinical presentation of several esophageal diseases is similar and the best method for a specific diagnosis of these patients has not been well defined. To evaluate the role of the polymerase chain reaction (PCR) in the etiologic definition of esophageal ulcers in HIV-1-infected patients, 96 esophageal biopsies from 79 HIV-1-infected patients were processed by PCR using specific primers for cytomegalovirus (CMV), herpes virus (HSV), human papilloma virus (HPV), HIV-1, Mycobacterium tuberculosis, Mycobacterium avium, Mycobacterium intracellulare, Treponema pallidum, and Haemophilus ducreyi. The PCR results were compared to the histopathologic results. Seventy-nine patients were studied (mean age: 34 years; 62% men; median CD4 + T cell = 103.59 cells/microl (range 1-795.2 cells/microl). The most common endoscopic findings were as follows: esophageal candidiasis (37.1%), esophageal ulcers (24.7%), esophagitis (11.2%), and lugol-negative areas (10.1%). The histopathologic findings in the esophageal ulcers (22 biopsies) were non-specific inflammation (31.8%), HSV (36.4%), Candida (13.6%), CMV (13.6%), or HPV disease (4.5%). In the esophageal ulcer biopsies, the PCR results were negative in 27.6% of cases, and positive for HIV (65.5%), CMV (31%), HPV (20.7%), HSV (10.3%), and H. ducreyi (6.9%). The histopathologic examination did not identify a pathogen or identified only Candida in 15 biopsies of esophageal ulcers. PCR was positive in ten (66.7%) and negative in five (33.3%) of these biopsies (idiopathic ulcers). PCR detected: HIV (53.3%), CMV (20%), HPV (13.3%), and H. ducreyi (6,7%). PCR detected more etiologic agents in esophageal ulcers than histopathology and was able to detect unusual pathogens. On the other hand, sometimes more than one pathogen was detected in the esophageal ulcers, making it difficult to reach an accurate diagnosis. This finding indicates the need for more studies to evaluate the benefit of this method in the routine evaluation of esophageal ulcer biopsies in HIV-1-infected patients.