The variability of the inverse solution provided by electrocardiographic imaging (ECGI) is largely unknown when comparing different leadsets or (similar) beats. In four patients, we compared activation times (ATs), recovery times (RTs), and correlation coefficients during QRS complex and STT segment between: 1) consecutive sinus beats within one leadset, and 2) multiple beats for two leadsets. Furthermore, reasons behind differences in RT were investigated. Zero-th order Tikhonov regularization was used to reconstruct ventricular epicardial potentials. A spatiotemporal estimation method was then used to determine the ATs and RTs from the reconstructed epicardial electrograms. Inter-leadset differences were generally low for ATs, but exceeded intra-leadset beat-to-beat variations. RTs, however, showed larger variation independent of leadset. Differences in RTs between beats or leadsets could partially be explained by low T-wave amplitudes and high levels of noise, which suggests that RT determination may require more advanced methods in these cases. These findings increase our understanding of the consequences of electrode placement for the inverse solution, as well as our understanding of the complexities of recovery time estimation in ECGI.