Body-surface dominant frequency (DF) mapping has been proposed as a technique for non-invasively identifying high-frequency cardiac sources during fibrillation. However, previous studies indicate that volume conduction could distort the spectrum of body-surface cardiac signals and hence, affect body-surface DF maps. In this study, we analyze the effects of volume conduction on the spectrum of cardiac signals in a realistic computer model of the canine ventricles. We simulate complex cardiac dynamics on the ventricular model and analyze the dependence of the bandwidth (BW) of simulated unipolar cardiac signals on the spatial resolution of the corresponding unipolar electrode, which we quantify with the lead equivalent volume (LEV). Our analysis shows that the BW decreases for increasing LEV values and saturates for high LEV values. Our results also indicate that the LEV saturation value is low for low degrees of spatiotemporal correlation. We conclude that the spectral effects of volume conduction might limit our ability to accurately identify high-frequency sources in body-surface DF maps during cardiac fibrillation.