1. The proposed mechanism of cellular plasticity underlying classical conditioning of Hermissenda involves Ca2+ influx through voltage-activated channels. This influx triggers several molecular cascades and leads to the phosphorylation of K+ channels in identified photoreceptors. We studied Ca2+ currents from isolated photoreceptors of Hermissenda with the whole cell patch-clamp technique. Two distinct Ca2+ currents were identified in isolated photoreceptors on the basis of differences in their voltage dependence, kinetics, and pharmacology. 2. One Ca2+ current was transient (ICa(t)), with a fast onset (approximately 5 ms), activated at -50 mV from a holding potential of -90 mV, and peaked at 0 mV. The second Ca2+ current, designated as sustained (ICa(s)), exhibited a delayed time-to-peak, activated at -30 mV, and reached maximum at 30 mV. 3. Steady-state activation curves for both currents were generated from normalized currents and fitted with the Boltzmann function; estimates of half-activation voltages for ICa(t) were -38.8 +/- 6.7 mV (mean +/- SD; n = 9) and 3.2 +/- 8.2 mV for ICa(s) (n = 11) with maximum slopes of 8.9 +/- 1.6 mV (n = 9) and 11.0 +/- 2.4 mV (n = 11). 4. The inactivation of ICa(s) was slow (time constants > 3 s) whereas ICa(t) inactivated rapidly (time constant of inactivation at various voltages; 75-600 ms). 5. Ni2+ (0.8 mM), Gd3+ (0.5 mM), and amiloride (10 microM) produced a reversible block of ICa(t) without affecting ICa(s). omega-Conotoxin GVIA (10 nM) irreversibly blocked ICa(s) whereas nitrendipine (20 microM) produced a reversible block. 6. ICa(t) may be responsible for steady-state membrane potential oscillations. ICa(s) may contribute to the maintenance of the amplitude of the plateau phase of the generator potential.