In previous studies, we established methodology for reconstructing endocardial potentials, electrograms and isochrones from a non-contact intracavitary probe during a single beat. The probe was too large to be introduced percutaneously. Here we examine the possibility of similar mapping with a small multielectrode catheter that could be introduced percutaneously and does not expand inside the cavity. Cavity geometry and endocardial potentials were recorded in an isolated canine left ventricle. Simulated catheter probes were introduced into the cavity. Probe potentials were computed from the measured endocardial potentials and perturbed to include measurement noise, geometrical errors, and limited electrode density. Endocardial potentials were then reconstructed from the perturbed probe potentials and compared to the actual measured potentials. Of all probes simulated, a 3.0 mm (9F) catheter that assumes a curved geometry (e.g., a J shape) inside the cavity performed best (better than a larger 7.6 mm cylinder simulating an inflatable probe). Without bending, a straight cylindrical probe of the same size (9F, 3.0 mm) did not perform well. Sixty probe electrodes were needed for accurate reconstruction. The J-probe reconstruction was very robust in the presence of noise (10%) and of geometry errors (3 mm shift, 10° rotation). The results demonstrate the feasibility of accurate single-beat endocardial mapping using a 9F percutaneous multielectrode catheter that assumes a J shape in the cavity without the need for expansion (e.g., into a balloon or a “basket”). The robustness of the procedure to noise and geometrical errors suggests its applicability in the clinical EP laboratory and the possibility of determining probe position in vivo using current imaging modalities.