The frequency of indeterminate Western blot (WB) seroreactivities against HTLV-I "gag encoded proteins" only, and the use of low specific diagnostic WB criteria led to the overestimation of HTLV-I seroprevalence in initial studies in intertropical Africa and Papua New Guinea. In order to clarify the meaning of such seroreactivity, 98 blood samples of individuals from a high HTLV-I endemic area in Zaire, Central Africa were studied by a WB assay containing HTLV-I disrupted virions enriched with a gp 21 recombinant protein and a synthetic peptide from the gp 46 region (MTA-1), and by the polymerase chain reaction (PCR) with 3 primers pairs and 4 different HTLV-I and or HTLV-II-specific probes. These 98 samples were taken mainly from patients with neurological diseases and from their relatives. Using stringent WB criteria, 28 sera (29%) were considered as HTLV-I-positive, 3 as negative and 67 (68%) as indeterminate. A large proportion of these indeterminate sera would have been considered as HTLV-I-positive samples according to previous low specific WB diagnostic criteria. After PCR, 35 samples (36%) were considered as positive for the presence of HTLV-I proviral DNA. Out of the 67 WB seroindeterminate, 10 (15%) were found HTLV-I-positive by PCR. These 10 individuals exhibited in WB multiple band reactivity with p19 and/or p24 (7 cases of both) associated in 6 cases with rgp 21, but never with MTA-1. No samples were found PCR-positive for HTLV-II despite the findings of 11 sera suggestive of HTLV-II by WB.(ABSTRACT TRUNCATED AT 250 WORDS)