Spatial organization has been observed during episodes of ventricular fibrillation (VF) by recording epicardial unipolar electrograms on a grid of electrodes. In such studies, the choice of spacing between electrodes is an important decision, affecting the resolution and the size of the domain to be studied. A basic tenet of sampling theory, the Nyquist criterion, states that an electrode spacing smaller than half the smallest significant wavelength is required to capture the important details of a spatially sampled process. In this paper, we suggest a method to choose a practical interelectrode spacing by examining wavenumber power spectra of high-resolution VF data recorded from a square 11 x 11 array of electrodes spaced 0.28 mm apart. The plaque was sutured on the epicardium near the left ventricular apex in seven anesthetized pigs. VF was induced with ac simulation. Unipolar extracellular electrograms were simultaneously recorded from each channel for 2 s after the induction of VF. Each signal was sampled in time at 1000 Hz. Wavenumber power spectra were calculated for 100 ms segments using the zero-delay wavenumber spectrum method, for a total of 140 power spectra. All spectra had dominant peaks at the origin and fell off rapidly with increasing wavenumber (decreasing wavelength). In all the spectra, every wavelength shorter than 1.4 mm contributed insignificant power. Furthermore, in 134 of 140 spectra (96%), insignificant power levels were associated with every wavelength shorter than 2.8 mm. These results suggest that, for unipolar extracellular electrodes, an intersensor spacing on the order of 1 mm is appropriate to study organization during early VF.