There are many hypotheses of relationships, and also of speciation processes, in North American freshwater fishes, although, to date, there have been no direct tests of whether there is evidence of ecological niche conservatism. In the present study, ecological niche modeling is used to look for evidence of ecological niche conservatism in six clades of freshwater fishes: the starheaded topminnows, sand darters, black basses, Notropis rubellus species group, Notropis longirostris species group, and the Hybopsis amblops species group. This is achieved by evaluating the reciprocal predictivity of distributional predictions based on ecological niche models developed for each individual taxon in a clade under the assumption that high reciprical predictivity between sister species can be taken as evidence of niche conservatism. Omission percentages, total and average commission, and the area under the curve in a receiver operating characteristic analysis, where calculated, are used to evaluate predictive ability. Occurrence data for each species were subset into a training and independent validation data set where possible. Across all clades and species, models predicted the validation data for a given species well. Ecological niche conservatism was found generally across the starheaded topminnows, the sand darters, and the N. longirostris species group. There was some inter-predictivity within the N. rubellus group, but almost no inter-predictivity within the black basses, indicating a lack of conservatism. These results demonstrate that ecological niches generally act as stable constraints on freshwater fish distributions in North America. © 2009 The Linnean Society of London, Biological Journal of the Linnean Society, 2009, 96, 282–295.