The objective of this study was to assess the factors implicated in an increased or decreased risk of pneumonia, with particular attention to the response to highly active antiretroviral therapy (HAART) and the effect of the polysaccharide 23-valent pneumococcal vaccination in 300 human immunodeficiency virus (HIV)-infected adults followed-up for a median of 35.6 months. Pneumococcal pneumonia occurred in 12 patients and all bacterial pneumonia (pneumonia caused by Streptococcus pneumoniae or other bacteria, as well as those with negative cultures but presumably bacterial in origin) in 40 patients. In the univariate analysis, immunodepressed patients (defined as those with less than 200 CD4+ T cell/μl), those without immunological response to HAART (defined as an increase of 25% of CD4+ T lymphocyte count), patients with previous admissions to hospital and those with cotrimoxazole or Mycobacterium avium intracellulare prophylaxis showed a higher incidence of both pneumococcal and all bacterial pneumonia. Multivariate analysis demonstrated that the presence of pneumococcal pneumonia was associated with a CD4+ lymphocyte count at the time of HIV diagnosis <200 cells/μl. The multivariate model that was more valid for prediction of all bacterial pneumonia included a CD4+ T cell count <200 cells/μl and absence of immunological response to HAART. Only in patients with a baseline CD4+ T cell count lower than 200/μl and immunological response to HAART, a near significant lower incidence of all bacterial pneumonia was observed after vaccination. Thus, these results do not support an important additional protective effect of 23-valent pneumococcal vaccine in HIV-patients with immunological response to HAART. J. Med. Virol. 72:517–524, 2004. © 2004 Wiley-Liss, Inc.