A recent analysis of amino acid sequence data (Graur et al.) suggested that the mammalian order Rodentia is polyphyletic, in contrast to most morphological data, which support rodent monophyly. At issue is whether the hystricognath rodents, such as the guinea pig, represent an independent evolutionary lineage within mammals, separate from the sciurognath rodents. To resolve this problem, we sequenced a region (2,645 bp) of the mitochondrial genome of the guinea pig containing the complete 12S ribosomal RNA, 16S ribosomal RNA, and transfer RNA(VAL) genes for comparison with the available sciurognath and other mammalian sequences. Several methods of analysis and statistical tests of the data all show strong support for rodent monophyly (91%-98% bootstrap probability, or BP). Calibration with the mammalian fossil record suggests a Cretaceous date (107 mya) for the divergence of sciurognaths and hystricognaths. An older date (38 mya) for the controversial Mus-Rattus divergence also is supported by these data. Our neighbor-joining analyses of all available sequence data (25 genes) confirm that some individual genes support rodent polyphyly but that tandem analysis of all data does not. We propose that the conflicting results are due to several compounding factors. The unique biochemical properties of some hystricognath metabolic proteins, largely responsible for generating this controversy, may have a single explanation: a cascade effect resulting from inactivation of the zinc-binding abilities of insulin. After excluding six genes possibly affected by insulin inactivation, analyses of all available sequence data (7,117 nucleotide sites, 3,099 amino acid sites) resulted in strong support for rodent monophyly (94% BP for DNA sequences, 90% for protein sequences), which lends support to the insulin-cascade hypothesis.