The taxonomy of the African dwarf crocodile (genus Osteolaemus) has been disputed since a novel morphotype was discovered in the early 20th Century. Because this poorly-known reptile is widely hunted throughout the forests of Central and West Africa, resolving the existence and extent of taxonomic units has important management and conservation implications. Lack of molecular data from individuals of known origin and historical disagreement on diagnostic morphological characters have hindered attempts to settle one of the most important taxonomic questions in the Crocodylia. In an effort to clarify the evolutionary relationships among dwarf crocodiles, we sequenced three mitochondrial and two nuclear genes using a large sample of dwarf crocodiles from known localities across major drainage basins of forested Africa. Concordant results from Bayesian, maximum likelihood, maximum parsimony and population aggregation analytical methods support a previously recognized division of the dwarf crocodile into a Congo Basin form (O. osborni) and a West African form (Osteolaemus tetraspis), but also reveal a third diagnosable lineage from West Africa warranting recognition as an separate taxonomic unit. Corrected genetic distances between geographic regions ranged from 0.2% to 0.6% in nuclear fragments and 10.0 to 16.2% in mitochondrial COI. Population aggregation, using fixed and alternate character (nucleotide) states to cluster or divide populations, recovered 232 such molecular characters in 4286 bp of sequence data and unambiguously aggregated populations into their respective geographic clade. Several previously recognized morphological differences coincide with our molecular analysis to distinguish Congo Basin crocodiles from the Ogooué Basin and West Africa. Discrete morphological characters have not yet been documented between the latter two regions, suggesting further work is needed or molecular data may be required to recognize taxonomic divisions in cases where putative species are morphologically cryptic. This study highlights the importance of using widespread taxon sampling and a multiple evidence approach to diagnose species boundaries and reveal cryptic diversity.