Exposure of the perfused rat liver to a perfusate made hyperosmotic by the presence of 200 mmol l-1 glucose led, as expected, to marked, transient hepatocellular shrinkage followed by volume-regulatory net K+ uptake. However, even after this volume-regulatory K+ uptake had ceased, the liver cells are still slightly shrunken. Withdrawal of glucose from the perfusate resulted in marked transient cell swelling, net K+ release from the liver and restoration of cell volume. However, when the Krebs-Henseleit perfusate was made hyperosmotic by the presence of urea (20-300 mM), there was no immediate decrease in liver mass, yet a slight and persistent cell shrinkage developing 2 min after the onset of exposure to urea. Surprisingly, urea induced concentration-dependent net K+ efflux from the liver and removal of urea net K+ reuptake from the inflowing perfusate. The urea (200 mM)-induced net K+ release resembled that observed following a lowering of the influent [NaCl]: making the perfusate hypoosmotic (245 mosmol l-1, by reducing influent [NaCl] by 30 mM) gave roughly the same K+ response as hyperosmotic exposure (505 mosmol/l) resulting from the presence of 200 mM urea. The urea-induced K+ efflux was not inhibited in the presence of ouabain (1 mM), or in Ca(++)-free perfusion, but was modified in the presence of quinidine (1 mM) or Ba++ (1 mM). The direction in which the liver was perfused had no effect on the urea-induced net K+ release.(ABSTRACT TRUNCATED AT 250 WORDS)