Two models have been proposed to explain facilitation of the L-type calcium current (ICa-L). A positive feedback model proposes that calcium released during a conditioning pulse (I1) facilitates the subsequent pulse (I2) via calmodulin/calmodulin kinase II (CaMKII) mechanisms. The negative feedback model proposes that the calcium release of each pulse feeds back on itself via calcium-dependent inactivation. The relative physiological roles were evaluated in rat ventricular myocytes. Paired pulses (450 ms interpulse interval) elicited facilitation (I2 of 872 +/- 145 versus I1 of 777 +/- 132 pA, P < 0.01). Inactivation time (T0.37) was prolonged for I2 versus I1 (22 +/- 2 and 16 +/- 2 ms, P > 0.01). Evidence for the negative feedback mechanism includes: (a) ryanodine (0.3 mM ) eliminated facilitation, surprisingly by increasing the amplitude of I1 more than that of I2 (1039 +/- 216 and 977 +/- 186 pA) and eliminated the difference in T0.37 between I2 and I1 (33.1 +/- 4.5 versus 32.5 +/- 4.6 ms); (b) an outward I2, which does not trigger sarcoplasmic reticulum (SR) Ca2+ release, eliminated facilitation even when it was conditioned by an inward I1; (c) facilitation decayed as the I1-I2 interval lengthened (time constant (tau) = 16.9 +/- 1.4 s); (d) thapsigargin (0.1 microM ) slowed this decay (tau = 43.8 +/- 11.7 s) whereas isoproterenol accelerated it (tau = 5.6 +/- 1.4 s, P < 0.01) and T0.37 paralleled this decay; and (e) the magnitude of ICa-L was negatively correlated with the sodium-calcium exchange current (INa/Ca) elicited by the SR-Ca2+ release. In conclusion, Ca2+-dependent inactivation of ICa-L is the major mechanism underlying facilitation.