A total of 45 strains of Mycobacterium avium from 31 human immunodeficiency virus (HIV)-positive patients in the French Caribbean islands and Guiana were subjected to DNA fingerprinting using a recently described consensus IS1245 restriction fragment length polymorphism (RFLP) method, and pulsed-field gel electrophoresis (PFGE). The IS1245-RFLP resulted in three distinct clusters composed of three 27-banded isolates from two patients (cluster A), nine two-banded isolates from six patients (cluster B), and three 20-banded isolates from three patients (cluster C). PFGE results obtained after XbaI and DraI digestions gave similar clustering results irrespective of the enzyme used, and confirmed the molecular clonality for high IS1245 copy number isolates (clusters A and C). However, PFGE further discriminated the low IS1245 copy number cluster B into two distinct subclusters: subcluster I containing six isolates from four patients during the same time period from a single hospital in Guadeloupe, and subcluster IV composed of four isolates from two patients, out of which three isolates were from a single patient (patient 19). Interestingly, in the latter case, PFGE grouped together all three isolates from patient 19 despite the fact that IS1245 fingerprinting permitted grouping only two of the three isolates (the remaining unclustered isolate contained two additional bands of 3.5 and 5 kbp, and was initially considered as evidence of polyclonal infection). A combined numerical analysis of the IS1245-RFLP and PFGE results corroborated the existence of four instead of three clusters. A comparison of IS1245-RFLP versus PFGE results suggested that the standardized RFLP procedure is compatible with PFGE only for M. avium isolates containing > or = 5 copies of IS1245. Consequently, the typing results for low IS1245 copy number isolates (31% of isolates in this study) should be reconsidered for secondary typing using PFGE. Lastly, the absence of a predominant genotype of M. avium infecting HIV-positive patients over a 5-year period in this tropical region argues in favor of a lack of a privileged ecological niche for M. avium, and instead suggests that microepidemics of M. avium may prevail during limited periods of time.