Melanoma development and progression is thought to be the result of a multi-step accumulation of genetic damage, with loss of heterozygosity in chromosome 9p (MTS1) frequently described. In addition, chromosome 10q allelic loss has been reported, implicating the tumor suppressor gene PTEN/MMAC1 on 10q23.3. The MXI1 gene at 10q24-25 is another candidate tumor suppressor that has only rarely been studied in melanomas, with conflicting results. We used microdissection-based genotyping to investigate 29 melanomas from 20 patients for loss of heterozygosity in intragenic and flanking microsatellite markers for this latter gene. Concurrently, the MTS1 gene was similarly studied using two flanking microsatellites. Fifty-four percent (15 of 28) of the informative cases showed loss of heterozygosity for one or both MXI1 markers, as compared with 67% (16 of 24) of the informative cases for MTS1. MXI1 allelic loss was seen more frequently in recurrent/metastatic tumors (59%), as compared with in primary (33%) lesions. Eighty percent of the primary tumors showed loss of heterozygosity for MTS1, as well as 63% of recurrent/metastatic ones. We studied more than one tumor in eight patients, with those from three patients showing discordant genetic patterns. One patient showed a metastatic tumor with allelic loss for MXI1 that was not identified in the primary melanoma or a local recurrence. The other two patients showed clonal heterogeneity in MXI1 at synchronous and metachronous metastatic foci. These findings support MXI1 as a putative tumor suppressor gene involved in conventional melanoma progression. Genetic heterogeneity seen in different metastases from the same primary suggests a nonlinear pattern of chromosomal damage, with the development of multiple clones within the primary tumor, each acquiring its own metastatic potential.