Ribotyping has previously been used for epidemiological studies of Burkholderia pseudomallei (previously Pseudomonas pseudomallei). We show here that random amplified polymorphic DNA (RAPD) analysis allows subdivision of strains of the same ribotype. With five different primers, no two epidemiologically unrelated isolates of any single ribotype in this study of 102 isolates from humans, goats, cats, and soil had identical RAPD patterns. Conversely, RAPD analysis showed clonality for isolates from each of two animal outbreaks of melioidosis and from a nontropical focus of animal and human melioidosis spanning 25 years. Some soil isolates were identical to epidemiologically related animal and human isolates as determined by RAPD typing. There was no evidence that the clinical outcome of melioidosis was related to RAPD patterns.