The dependence of KATP channel activity on [ATP] has been examined in isolated membrane patches from rat ventricular myocytes. The steady-state [ATP] dependence of channel open probability could be described by a sigmoidal relationship with the ki ([ATP] causing half-maximal inhibition of open probability) = 25 microM and Hill coefficient of 2. Description of channel open- and closed-time distributions required at least 2, and 3, time constants, respectively. Long open-channel lifetimes decreased with [ATP]; unconditional mean channel closed-times increased with [ATP]. Step decrease (jump) in bathing [ATP] resulted in a delay (of up to hundreds of milliseconds) followed by a pseudo-exponential rise of current (with a time constant of up to hundreds of milliseconds). The time course of channel current after changes of [ATP] (or the ATP-analogue AMP-PNP) was shown to be predominantly determined by the time course of diffusion into the tip of the electrode and to the membrane. This time course of diffusion of ATP into the pipette tip had to be taken into account when analyzing the current response to [ATP] steps. Several possible kinetic models of the ATP-dependent regulation of channel activity were considered. Adequate explanation of the data required a model with sequential ATP-binding sites. The model can account for the time course of channel opening after steps of [ATP], as well as for the steady-state dependence of P0 on [ATP]. The model predicts [ATP]-dependent closed and open lifetimes as were observed experimentally.