The skin overlying the cleithrum bone of freshwater-acclimated rainbow trout contains numerous mitochondria-rich (MR) cells, as detected by DASPEI fluorescence. This tissue was mounted in vitro in an Ussing-style chamber with fresh water on the mucosal surface and saline supplemented with bovine serum albumin on the serosal surface. The preparation developed a high transepithelial resistance and a small transepithelial potential (Vt), positive on the serosal side. Radioisotopic flux measurements indicated that the preparation actively transported Ca2+ from the mucosal to the serosal surface, as assessed by the Ussing flux ratio criterion. Ca2+ transport was positively correlated with MR cell density. Cortisol pretreatment in vivo reduced MR cell density and increased Vt but did not significantly alter Ca2+ fluxes. Ca2+ transport was unaffected by adrenergic agonists (10(-5) mol l-1 adrenaline, clonidine, isoprenaline) or cyclic AMP stimulants (10(-3) mol l-1 dibutyryl cyclic adenosine monophosphate, db-cAMP, plus 10(-4) mol l-1 isobutylmethylxanthine, IBMX) applied to the serosal surface. The Ca2+ ionophore ionomycin (1 x 10(-6)-3.2 x 10(-6) mol l-1 on the mucosal surface) increased both unidirectional Ca2+ fluxes and caused Ca2+ to accumulate within the epithelium. Lanthanum (10(-4) mol l-1) did not inhibit unidirectional Ca2+ fluxes, but apparently displaced Ca2+ from binding sites on the mucosal surface. Unlike Ca2+, movements of Na+ and Cl- across the epithelium were passive, as assessed by the flux ratio criterion, and neither adrenaline nor db-cAMP plus IBMX had any effect on Na+ or Cl- fluxes or electrical properties. These results indicate that ion transport across the skin mediated by MR cells ('chloride cells') contributes to Ca2+ but not to NaCl balance in freshwater trout.