1. Monophasic action potentials (MAPs) were recorded simultaneously from the endocardium and the epicardium in open-chested dogs (n = 16) during bolus intravenous injections of potassium chloride at 0.1 mmol/kg and 0.2 mmol/kg. MAP duration was measured at 90% repolarization. Local conduction was measured as the delay between the pacing artefact and upstroke of the MAP. 2. The animals were anaesthetized with alpha-chloralose and urethane. Heart block was created by atrioventricular nodal ablation with 0.1% (v/v) formalin. Infra-nodal pacing was established. 3. Bolus injections of potassium chloride at 0.1 mmol/kg shortened MAP duration. The effect was maximal at 15 s after the injection [epicardium: 171.5 +/- 3.0 (control) and 148.9 +/- 5.2 ms at 15 s (P less than 0.001); endocardium: 178.7 +/- 4.4 (control) and 165.6 +/- 5.3 ms (P less than 0.001)]. The shortening on the epicardium was significantly greater than on the endocardium (P less than 0.05). 4. Local conduction time showed no significant change [epicardium: 37.8 +/- 1.8 (control) and 41.2 +/- 2.4 at 15 s; endocardium: 41.1 +/- 1.6 (control) and 45.0 +/- 3.4 at 15 s]. 5. After beta-adrenoceptor blockade, MAP duration shortened, exhibiting a maximum effect at 20 s after a bolus injection of potassium chloride at 0.1 mmol/kg [epicardium: 184.7 +/- 7.1 (control) and 130.7 +/- 9.0 ms at 20 s (P less than 0.001); endocardium 192.0 +/- 7.4 (control) and 148.6 +/- 10.8 ms at 20 s (P less than 0.001)]. The greater shortening observed on the epicardial surface did not reach statistical significance.(ABSTRACT TRUNCATED AT 250 WORDS)