1. The three-micro-electrode voltage clamp method (Adrian, Chandler & Hodgkin, 1970) was adapted for the study of regenerative inward currents in cardiac muscle. The adequacy of measurements of slow inward current in cardiac Purkinje fibres was assessed. 2. Membrane current density is reported simultaneously by total applied current (IT), along with a longitudinal voltage difference signal (delta V), recorded between two intracellular micro-electrodes. 3. Non-linear cable calculations show that delta V is a more faithful measure of membrane current density than IT as peak inward current increases. Quantitative agreement between delta V and IT only occurs when both signals report the membrane characteristics that would be obtained with an ideal longitudinal space clamp. 4. Agreement between delta V and IT is thus a useful criterion for satisfactory experimental measurements which we applied to the slow inward current. This component was elicited by depolarizing steps from a holding potential near =45 mV in the presence of tetrodotoxin. The IT signal was compared directly with delta V/R, where R is an effective longitudinal resistance that was experimentally determined. 5. delta V/R and IT showed very good agreement in both peak amplitude and time course at all potentials studied. 6. Radial non-uniformity during the measured peak slow inward current was estimated by calculations assuming clefts 200 A wide with a uniform distribution of ionic channels. The calculated voltage span from surface to centre was always less than 5 mV, and the measured I-V characteristics showed little distortion. 7. In another check, I-V characteristics and slow response membrane action potentials were compared. The measured peak current showed good agreement with the product (total preparation capacitance) x (rate of rise). 8. The experimental and theoretical analysis suggest that the measurements of slow inward current are a good approximation to genuine membrane properties.