The source of thiaminase in the Great Lakes food web remains unknown. Biochemical characterization of the thiaminase I activities observed in forage fish was undertaken to provide insights into potential thiaminase sources and to optimize catalytic assay conditions. We measured the thiaminase I activities of crude extracts from five forage fish species and one strain of Paenibacillus thiaminolyticus over a range of pH values. The clupeids, alewife Alosa pseudoharengus and gizzard shad Dorosoma cepedianum, had very similar thiaminase I pH dependencies, with optimal activity ranges (> or = 90% of maximum activity) between pH 4.6 and 5.5. Rainbow smelt Osmerus mordax and spottail shiner Notropis hudsonius had optimal activity ranges between pH 5.5-6.6. The thiaminase I activity pH dependence profile of P. thiaminolyticus had an optimal activity range between pH 5.4 and 6.3, which was similar to the optimal range for rainbow smelt and spottail shiners. Incubation of P. thiaminolyticus extracts with extracts from bloater Coregonus hoyi (normally, bloaters have little or no detectable thiaminase I activity) did not significantly alter the pH dependence profile of P. thiaminolyticus-derived thiaminase I, such that it continued to resemble that of the rainbow smelt and spottail shiner, with an apparent optimal activity range between pH 5.7 and 6.6. These data are consistent with the hypothesis of a bacterial source for thiaminase I in the nonclupeid species of forage fish; however, the data also suggest different sources of thiaminase I enzymes in the clupeid species.