Individual myocytes were isolated from bullfrog atrium by enzymatic and mechanical dispersion, and a one-microelectrode voltage clamp was used to record the slow outward K+ currents. In normal [K+]o (2.5 mM), the slow outward current tails reverse between -95 and -100 mV. This finding, and the observed 51-mV shift of Erev/10-fold change in [K+]o, strongly suggest that the "delayed rectifier" in bullfrog atrial cells is a K+ current. This current, IK, plays an important role in initiating repolarization, and it is distinct from the quasi-instantaneous, inwardly rectifying background current, IK. In atrial cells, IK does not exhibit inactivation, and very long depolarizing clamp steps (20 s) can be applied without producing extracellular K+ accumulation. The possibility of [K+]o accumulation contributing to these slow outward current changes was assessed by (a) comparing reversal potentials measured after short (2 s) and very long (15 s) activating prepulses, and (b) studying the kinetics of IK at various holding potentials and after systematically altering [K+]o. In the absence of [K+]o accumulation, the steady state activation curve (n infinity) and fully activated current-voltage (I-V) relation can be obtained directly. The threshold of the n infinity curve is near -50 mV, and it approaches a maximum at +20 mV; the half-activation point is approximately -16 mV. The fully activated I-V curve of IK is approximately linear in the range -40 to +30 mV. Semilog plots of the current tails show that each tail is a single-exponential function, which suggests that only one Hodgkin-Huxley conductance underlies this slow outward current. Quantitative analysis of the time course of onset of IK and of the corresponding envelope of tails demonstrate that the activation variable, n, must be raised to the second power to fit the sigmoid onset accurately. The voltage dependence of the kinetics of IK was studied by recording and curve-fitting activating and deactivating (tail) currents. The resulting 1/tau n curve is U-shaped and somewhat asymmetric; IK exhibits strong voltage dependence in the diastolic range of potentials. Changes in the [Ca2+]o in the superfusing Ringer's, and/or addition of La3+ to block the transmembrane Ca2+ current, show that the time course and magnitude of IK are not significantly modulated by transmembrane Ca2+ movements, i.e., by ICa. These experimentally measured voltage- and time-dependent descriptors of IK strongly suggest an important functional role for IK in atrial tissue: it initiates repolarization and can be an important determinant of rate-induced changes in action potential duration.