Ammonia excretion rates of channel catfish, Ictalurus punctatus, little skate (Raja erinacea), and blue crab (Callinectes sapidus) were measured in experimental regimes which permitted simultaneous assessment of the partial pressure gradients for nonionized NH3 and the chemical concentration gradients of NH4+. Under conditions of low external ammonia, the average ammonia excretion was +295 microM kg-1 h-1 for catfish, +149 microM kg-1 h-1 for blue crabs, and +59 microM kg-1 h1 for skates with partial pressure gradients of +72.5 mu Torr, +413 mu Torr, and +24.4 mu Torr, respectively; and [NH4+] gradients of +189 microM l-1, +643 microM l-1, and +107 microM l-1 (positive indicating greater from inside to medium). When the external ammonia was increased to 1.15 mM l-1, both gradients were reversed, and the net ammonia movement was initially from the external water into all three species. In the catfish the inward movement ceased, however, and ammonia excretion eventually resumed in the face of reversed gradients of both NH3 partial pressure and [NH4+]. Unidirectional Na+ influx, indicative of a Na+/NH4+ exchange, did not increase. The ammonia data, changes in titratable acidity, and net apparent H+ efflux were all consistent with a linked extrusion of internal NH4+ for external H+. Incorporation of such an exchange into a computer simulation model of the ammonia equilibrium and exchange system duplicated the experimental data. Other hypotheses failed to match experimental data, or failed to predict internal ammonia levels lower than outside. In the crab, internal ammonia levels rose rapidly to concentrations and partial pressures above the external medium until excretion was reestablished, with no evidence of maintenance of a reversed gradient. In the skate, internal concentrations rose appreciably in the first hour and continued to rise for 6-8 h, with no resumption of ammonia excretion. The interspecies differences appear to be due at least partly to differences in ammonia permeability of the gills.