1. Membrane currents and changes in [Ca2+]i attributable to the operation of an electrogenic Na-Ca exchange mechanism were recorded in single isolated guinea-pig ventricular myocytes under voltage clamp and internal perfusion with the Ca2+ indicator Fura-2. 2. Ionic currents that interfere with the measurement of Na-Ca exchange current were blocked through the use of caesium (Cs+), verapamil and tetrodotoxin (TTX). Entry of Ca2+ through surface membrane Ca2+ channels and release of Ca2+ from sarcoplasmic reticulum were blocked with verapamil and ryanodine, respectively. 3. In the presence of the blockers listed above, depolarization to positive membrane potentials elicited slow increases in [Ca2+]i and, after an instantaneous increase, a declining outward current. Repolarization elicited a decline in [Ca2+]i and, after an instantaneous increase, a declining inward current. The changes in [Ca2+]i and a major component of the current were abolished by nickel ions (Ni2+; 5 mM). 4. The reversal potential of the current abolished by Ni2+ (Ni2+-sensitive current) was determined at different levels of [Ca2+]i by ramp repolarizations from +80 to -80 mV (1-5 mV/ms). The reversal potential of the current increased linearly with log [Ca2+]i. As a result of the foregoing and other data, the Ni2+-sensitive current was taken to be Na-Ca exchange current (INaCa). 5. The relation between INaCa and [Ca2+]i (less than 1 microM) at constant voltage over the range of -80 to +60 mV was approximately linear. No evidence of saturation could be found; small deviations from linearity at high [Ca2+]i were in the direction expected for a minor contribution from Ca2+-activated non-specific cation current (Ehara, Noma & Ono, 1988). 6. When measured at the same [Ca2+]i, the peak INaCa upon repolarization to -80 to -140 mV seemed to approach a limiting value at very negative potentials. 7. Over the range of +40 to +160 mV INaCa (measured soon after depolarization and thus at the same [Ca2+]i) increased exponentially with clamp-pulse potential. These pulses (to potentials up to +160 mV) elicited a slow rise in [Ca2+]i with the peak at the end of the pulse also increasing exponentially with pulse potential. 8. Inward membrane currents with properties similar to those described above were also recorded in association with physiological [Ca2+]i transients, when Ca2+ channels and the sarcoplasmic reticulum were not blocked. 9. Some of the results are not consistent with certain predictions of a sequential step model, or with those of a simultaneous step model in which the internal binding site for Ca2+ is saturated, or with those of a model based only on thermodynamics.(ABSTRACT TRUNCATED AT 250 WORDS)