Aim: To examine the pattern and cumulative curve of descriptions of freshwater fishes world‐wide, the geographical biases in the available information on that fauna, the relationship between species richness and geographical rarity of such fishes, as well as to assess the relative contributions of different environmental factors on these variables.Location: Global.Methods: ModestR was used to summarize the geographical distribution of freshwater fish species using information available from data‐based geographical records. The first‐order jackknife richness estimator was used to estimate the completeness of such data in all terrestrial 1‐degree cells world‐wide. An α‐shape procedure was used to build range maps capable of providing relatively accurate species richness and geographical rarity values for each grid cell. We also examined the explanatory capacity of a high number of environmental variables using multiple regressions and Support Vector Machine.Results: Cumulative species description curves show that a high number of species of freshwater fishes remain to be discovered. Completeness values indicate that only 199 one‐degree grid cells, mainly located in eastern North America and Europe, could be considered as having relatively accurate inventories. Range maps provide species richness values that are positively and significantly related to those resulting from the first‐order jackknife richness estimator. The relationship between species richness and geographical rarity is triangular, so that these species‐rich cells are those with a higher proportion of distributionally rare species. Species richness is predicted by climatic and/or productivity variables but geographical rarity is not.Main conclusions: In general, species‐rich tropical areas harbour a higher number of narrowly distributed species although comparatively species‐poor subtropical cells may also contain narrowly distributed species. Historical factors may help to explain the faunistic composition of these latter areas; a supposition also supported by the low predictive capacity of climatic and productivity variables on geographical rarity values.