A high frequency of false-negative anti-HTLV-I/II ELISA results has been reported by several authors. To verify the possible underestimate of the prevalence of HTLV-II infection in subjects infected by HIV-1, we used the PCR to investigate the presence of HTLV DNA in peripheral blood mononuclear cells (PBMCs) collected from a group of 67 HIV-1-positive anti-HTLV-I/II ELISA-negative individuals; the study population included 31 patients with HIV-1-related peripheral neuropathy (PN), 15 with non-Hodgkin lymphoma (NHL), and 23 without PN or NHL. Two subjects had both PN and NHL. All of the patients who were positive at PCR were investigated for the presence of serum anti-HTLV-I/II antibodies by means of Western blot (WB). Eighteen (26.9%) of the 67 anti-HTLV-I/II ELISA-negative patients had HTLV DNA in their PBMCs and WB-detectable serum antibodies directed against one or more HTLV antigens. The individuals affected by predominantly sensory polyneuropathy (PSP) had a significantly higher prevalence of HTLV DNA than the others. All of the patients in whom HTLV-I/HTLV-II discrimination was successful had HTLV-II, with the exception of one patient who was infected by HTLV-I. The present study confirms the possibility of HTLV infection in the absence of ELISA-detectable serum anti-HTLV-I/II antibodies, especially in the particular setting of HIV-1-infected individuals. Moreover, the fact that the prevalence of HTLV DNA was significantly higher in the subjects affected by predominantly sensory polyneuropathy further supports the possibility of an association between HIV-1-related PSP and HTLV-II.