The effect of methapyrilene (MP), a mitochondrial proliferator and presumed nongenotoxic carcinogen, has been examined in rodent liver by means of high-resolution two-dimensional electrophoretic analysis of total proteins. Using this approach, we have discovered protein modifications in rat liver resulting from 1 week MP treatment that suggest the involvement of a reactive drug metabolite. The restriction of these molecular charge modifications to mitochondrial proteins indicates that such a reactive metabolite must be generated and confined within the mitochondrion. Quantitative changes in numerous nonmitochondrial proteins are also observed. Following a 4-week recovery period, almost all the 1-week treatment changes are reversed, reestablishing a protein pattern close to that of the controls. At the end of a 10-week exposure, the mitochondrial protein modifications are increased and are accompanied by a variety of quantitative protein changes indicative of a large shift in gene expression and/or cell type composition. When a 4-week untreated recovery period follows the 10-week treatment, small quantitative changes persist. In the mouse, where MP appears not to induce mitochondrial proliferation or tumorigenesis, 1 week treatment nevertheless produces mitochondrial protein changes in vivo consistent with attack by a reactive metabolite, but at a level substantially lower than that seen in the rat. Features of the mitochondrial protein modification indicate that it is covalent, does not involve cysteine or tryptophan, and results from binding of a negatively charged adduct. The possibility that the putative reactive metabolite could also attack mitochondrial (but not nuclear) DNA suggests that MP could be genotoxic in an unconventional way. Detection of protein modification by two-dimensional gel analysis appears to offer a general method for the detection and characterization of processes generating reactive metabolites.