|Abstract||Tetraodontiform fishes (e.g., triggerfishes, boxfishes, pufferfishes, and giant ocean sunfishes) have long been recognized as a monophyletic group. Morphological analyses have resulted in conflicting hypotheses of relationships among the tetraodontiform families. Molecular data from the single-copy nuclear gene RAG1 and from two mitochondrial ribosomal genes, 12S and 16S, were used to test these morphology-based hypotheses. Total evidence (RAG1 + 12S + 16S), RAG1-only, and mitochondrial-only analyses were performed using both maximum parsimony and Bayesian criteria. Total evidence and RAG1-only analyses recover a monophyletic Tetraodontiformes. However, the relationships recovered within the order differ, and none completely conform to previous hypotheses. Analysis of mitochondrial data alone fails to recover a monophyletic Tetraodontiformes and therefore does not support any of the morphology-based topologies. The RAG1 data appear to give the best estimate of tetraodontiform phylogeny, resulting in many strongly supported nodes and showing a high degree of congruence between both parsimony and Bayesian analyses. All analyses recover every tetraodontiform family for which more than one representative is included as a strongly supported monophyletic group. Balistidae and Monacanthidae are recovered as sister groups with robust support in every analysis, and all analyses except the Bayesian analyses of the mitochondrial data alone recover a strongly supported sister-group relationship between Tetraodontidae and Diodontidae. Many of the intrafamilial relationships recovered from the molecular data presented here corroborate previous morphological hypotheses. Copyright 2004 Elsevier Inc. All rights reserved.