Because treatment with lithium salts may impair renal concentrating ability, we investigated the possibility of a direct effect of lithium ions on the permeability to water of the collecting duct epithelium. The coefficient of hydraulic conductivity (Lp) of isolated perfused rabbit cortical collecting tubules (CCT) was measured in the presence and absence of arginine-8-vasopressin (AVP), or 8-bromo (Br) cyclic AMP (cAMP) and/or lithium chloride (Li 10 mM). In the absence of AVP, Li in the lumen for 30 min failed to affect basal water permeability; however, in tubules preincubated with Li in the lumen for 80 min, basal water permeability was reduced to 30% of the value found in control tubules (P less than 0.01). In CCT incubated at 25 degrees C with Li in the lumen for 3 h, the hydroosmotic response to 2.5 microU X ml-1 AVP (Lp = 6.88 +/- 1.54 nl X cm-2 X s-1 X atm-1) was significantly lower than that in the control tubules (13.98 +/- 1.59, P less than 0.01); the inhibition was not reversible. When Li was present in the peritubular medium only, the hydroosmotic effect of AVP was not different from that of the controls. The hydroosmotic effect of 25 microU/ml AVP was investigated at 37 degrees C. CCT exposed to Li in the lumen had a 49% inhibition of peak Lp under AVP (Lp = 10.98 +/- 1.17) as compared with control tubules (Lp = 21.39 +/- 1.51; P less than 0.005). In contrast, the hydroosmotic response to 8-Br-cAMP was not affected by lithium. The results are compatible with the view that Li inhibits the action of AVP at the level of the regulating protein or the catalytic unit of the membrane adenylate cyclase and that the site of the interaction can be reached by lithium only from the cytoplasmic side. The Li-antidiuretic hormone (ADH) interaction found here may represent the earliest pathophysiological event underlying the renal concentrating defect observed after Li administration.