1. In isolated rat heart ventricular myocytes exposed to 2 mM‐cyanide in the presence of 10 mM‐2‐deoxyglucose (complete metabolic blockade), there is a time‐dependent increase in ATP‐sensitive potassium (KATP) channel activity. The increase in KATP channel activity accompanies the decline of twitch amplitude. Channel activation and decline of the twitch amplitude precede the development of a ‘rigor’ contracture. 2. We measured KATP channel activity in permeabilized cells using the open‐cell attached (O‐C‐A) patch configuration (by establishing a cell‐attached patch and then permeabilizing the cell by exposure to saponin). The apparent ATP dependence of KATP channel activity could be described by a sigmoid curve with ki, ATP (ATP concentration required for half‐maximum inhibition of channel activity) = 122 microM and H (Hill coefficient) = 1.225. 3. In the O‐C‐A patch configuration, 10 mM‐creatine phosphate (CrP) decreased the apparent ki, ATP from 122 microM to about 10 microM, and the maximal activity (in zero ATP) was decreased to about 30% of the maximal activity in the absence of CrP. 4. In isolated inside‐out (I‐O) patches, ATP inhibited KATP channel activity at much lower [ATP] than in the O‐C‐A patch configuration (ki, ATP = 25 microM, H = 2). CrP was without effect on I‐O patches. 5. These results are consistent with the hypothesis that the difference in the ATP dependence of KATP channel activity in the O‐C‐A and I‐O patch configurations arises because of ATP consumption in the O‐C‐A patch configuration. The results suggest that hydrolysis of ATP to ADP by endogenous ATPases leads to the development of gradients of [ATP] and [ADP] between the bath and the ‘inside’ of the open cell. By re‐phosphorylating ADP, CrP is able to dissipate these gradients, revealing the ‘true’ ATP dependence of channel activity, which is the same as that in the I‐O patch configuration. 6. In order to estimate the contribution of KATP channel activity to the rat cardiac action potential at different [ATP] we have made the following measurements. Using electrodes of resistance 2‐8 M omega the density of KATP channels was 10.3 +/‐ 0.1 channels per patch (n = 162).(ABSTRACT TRUNCATED AT 400 WORDS)