A prospective study was conducted at Bamrasnaradura Hospital, Nonthaburi Province, Thailand from November 11, 2002 to January 5, 2003. A total of 59 HIV/AIDS patients with interstitial infiltrates on chest radiographs were included in the study. The objectives of this study were to describe the clinical manifestations and determine the etiologies of interstitial pneumonitis, assess the short-term outcomes and determine the accuracy of the clinical diagnosis of the etiologies of interstitial pneumonitis in HIV/AIDS patients at Bamrasnaradura Hospital, Nonthaburi, Thailand. Tuberculosis was the most common diagnosis (44%), followed by Pneumocystis carinii pneumonia (25.4%), bacterial pneumonia (20.3%) and fungal pneumonia (10.2%). In tuberculosis, compared to other diagnoses, a mild cough (p = 0.031), pallor (p = 0.021), lymphadenopathy (p < 0.001), absence of skin lesions (p = 0.003), higher mean body temperature (p = 0.004) and an absence of dyspnoea on exertion (p = 0.042) were significant findings. On multivariate analysis, however, only an absence of skin lesions (p = 0.023) remained a statistically significant predictor of TB. In Pneumocystis carinii pneumonia compared to other diagnoses, dyspnea on exertion (p = 0.014), non-purulent sputum production (p = 0.047), a higher mean respiratory rate (p < 0.001), absence of lymphadenopathy (p < 0.001) and lack of purulent sputum (p = 0.030) were significant factors. By multivariate analysis, only an absence of lymphadenopathy were shown to be independently and statistically significantly associated (p = 0.040). In bacterial pneumonia, compared to other diagnoses, production of purulent sputum (p = 0.014), hemoptysis (p = 0.006), pallor (p = 0), skin lesions (p = 0.002) and a severe cough (p = 0.020) were significantly associated factors. On multivariate analysis, none of these factors were statistically significant. In fungal pneumonia, compared to other diagnoses, headache and papulonecrotic skin lesions were common findings, but no factor had a significant association. After four weeks, 59.3% of the patients were alive, 13.6% died and 27.1% were lost to follow-up. Among the alive patients 88.6% had clinically improved. On multivariate analysis, no factor was shown to be a statistically significant predictor of death. The cumulative survival after 28 days was highest among PCP patients, followed by bacterial pneumonia, tuberculosis and fungal pneumonia, but this difference was not statistically significant (p = 0.0453).