We have observed that serum potassium levels measured immediately before induction of anesthesia ("preinduction K+") are often lower than those measured 1-3 days preoperatively ("preoperative K+"). The purpose of this investigation was to determine, in two studies, the magnitude of this difference and to elucidate the mechanism by which this occurs. In the first study, preinduction K+ (3.6 +/- 0.4 mEq/L, mean +/- SD) was significantly lower than K+ levels measured during routine preoperative testing (4.4 +/- 0.4 mEq/L, n = 47, P less than 0.001). Twenty-three patients (49%) had preinduction K+ levels that were considered hypokalemic (less than or equal to 3.5 mEq/L), but 22 of these 23 patients had normal preoperative K+ levels. The second study tested the hypothesis that preinduction decreases in serum K+ are mediated by beta 2-adrenergic receptors. Preinduction K+ changes were determined in patients given a single preoperative dose of propranolol (beta 1/beta 2-antagonist), atenolol (beta 1-antagonist), or no beta-blocker (control). The difference between preoperative and preinduction serum K+ in patients receiving propranolol (0.1 +/- 0.4 mEq/L) was significantly attenuated (P less than 0.02) compared with the difference in control subjects (0.5 +/- 0.4 mEq/L), but was not significantly different from controls in patients pretreated with atenolol (0.3 +/- 0.4 mEq/L). These results demonstrate that serum K+ levels measured intraoperatively just before anesthetic induction are consistently and significantly less than those measured 1-3 days preoperatively. The ability of propranolol but not atenolol to block this change suggests that the acute decrease in K+ levels was due to beta 2-adrenergic receptor stimulation.(ABSTRACT TRUNCATED AT 250 WORDS)