The process by which some eukaryotic organelles, for example the endomembrane system, evolved without endosymbiotic input remains poorly understood. This problem largely arises because many major cellular systems predate the last common eukaryotic ancestor (LCEA) and thus do not provide examples of organellogenesis in progress. A model is emerging whereby gene duplication and divergence of multiple ''specificity-'' or ''identity-'' encoding proteins for the various endomembranous organelles produced the diversity of nonendosymbiotically derived cellular compartments present in modern eukaryotes. To address this possibility, we analyzed three molecular components of the endocytic membrane-trafficking machinery. Phylogenetic analyses of the endocytic syntaxins, Rab 5, and the -adaptins each reveal a pattern of ancestral, undifferentiated endocytic homologues in the LCEA. Subsequently, these undifferentiated progenitors independently duplicated in widely divergent lineages, convergently producing components with similar endocytic roles, e.g., 1 and 2-adaptin. In contrast, 3, 4, and all other adaptin complex subunits, as well as paralogues of the syntaxins and Rabs specific for the other membrane-trafficking organelles, all evolved before the LCEA. Thus, the process giving rise to the differentiated organelles of the endocytic system appears to have been interrupted by the major speciation event that produced the extant eukaryotic lineages. These results suggest that although many endocytic components evolved before the LCEA, other major features evolved independently and convergently after diversification into the primary eukaryotic supergroups. This finding provides an example of a basic cellular system that was simpler in the LCEA than in many extant eukaryotes and yields insight into nonendosymbiotic organelle evolution.adaptin ͉ Rab ͉ SNARE ͉ trafficking ͉ autogenous