Neuronal somata of Lymnaea stagnalis were internally perfused and voltage clamped using the suction pipette method. The cells were exposed to internal solutions buffered to various concentrations of Ca2+ while the cytoplasmic Ca2+ activity [( Ca2+]i) was monitored with a Ca2+ -sensitive micro-electrode. [Ca2+]i was usually about 10(-7) M when the cell was perfused with a solution buffered to any level of Ca2+ from 9 X 10(-7) to below 10(-8) M. With internal solutions buffered to 10(-6) M-Ca2+ or greater, [Ca2+]i increased rapidly and overshot the perfusate Ca2+ activity by up to two orders of magnitude. It was thus virtually impossible to hold [Ca2+]i steady at any levels other than about 10(-7) M or 10(-4) M using internal perfusion of simple ionic internal solutions. The excess Ca2+ which caused the overshoot of [Ca2+]i entered the cell from the external solution through Cd2+ -sensitive channels. Cd2+ in the external solution prevented or reversed the overshoot of [Ca2+]i and brought [Ca2+]i to near the perfusate level. ATP added to the internal solution also prevented [Ca2+]i from overshooting the perfusate level during perfusion with high-Ca2+ buffers. By monitoring [Ca2+]i with a Ca2+ -sensitive micro-electrode, we were able to estimate the relationship between [Ca2+]i and the Ca2+ current (ICa) measured under voltage clamp. ICa was completely blocked as [Ca2+]i was raised to 10(-6) M. We believe that the discrepancy between our data and other estimates of the ICa vs. [Ca2+]i relationship using internal perfusion of molluscan nerve cells results from the incorrect assumption that [Ca2+]i is controlled adequately during internal perfusion.