In order to analyze the efficiency of interphase FISH for the detection and monitoring of Ph+ cells in chronic myelogenous leukemia (CML) under interferon (IFN) treatment, the following experiments were performed: (1) 98 specimens derived from 32 patients were analyzed in parallel by dual-color FISH and by conventional chromosome analysis (CCA). A 300/200 kb BCR/ABL probe was used in all tests and a smaller 35.5/39 kb probe was tested in parallel in 22 BM samples; (2) 30 BM samples were prepared by direct harvest and by 24-h culture and were analyzed in parallel; (3) PB and BM samples obtained simultaneously from 11 patients were analyzed. The cut-off point for the recognition of BCR/ABL fusion was set at 2.4%, calculated as the mean percent of false positivity in 11 controls plus 3 s.d. A very close correlation was observed (r=0.994, r2=0.988, P < 0.0001) between the percentages of Ph+ cells as assessed by CCA and by interphase FISH in 98 samples (26 at diagnosis). There was a moderate overestimation of the frequency of Ph+ cells by FISH with respect to CCA, that was more evident at low-to-medium values of Ph positivity. Seven specimens without Ph+ metaphases (17-50 cells analyzed) were shown to carry 2.5-8% interphase cells with BCR/ABL fusion. Similar percentages of BCR/ABL+ nuclei were recorded in 22 samples hybridized using the 300/200 kb and the 35.5/39 kb probe-sets (variation range: 0-5%, mean 2.3%). A very good correlation between the frequency of Ph+ interphase cells was observed when analyzing in parallel BM preparations after direct harvest and after 24-h culture. Underestimation of the percentage of BCR/ABL+ cells was noted to occur in 2/11 PB samples, compared to BM samples, the remaining nine cases showing superimposable results at either sites. We arrived at the following conclusions: (1) dual-color FISH enables an accurate detection and monitoring of the size of the Ph-positive clone in CML at diagnosis and after IFN-therapy; (2) FISH is more accurate than CCA, especially at low levels of Ph-positive cells; (3) testing of directly harvested BM samples is feasible and accurate, giving the opportunity to perform centralized FISH analysis in the context of multicentre trials; (4) the percentage of BCR/ABL+ PB cells usually, though not invariably, reflects the frequency of mutated cells in the BM.