Electrogram analysis is important in clinical and experimental settings. Activation recovery interval (ARI) has been used to measure ventricular action potential duration (APD) but its suitability for the atria has not been addressed. Mapping of atrial repolarization may be especially important during nerve stimulation since large heterogenous APD changes may manifest. This study assessed the utility of estimating APD in the atria using electrograms. A computer model of the atria was used to compute electrograms. Two different atrial waveforms were used, as well as two ventricular. APD was modulated with an acetylcholine- (ACh) dependent potassium channel and varying the spatial ACh distribution. ARI was computed, as well as the area under the repolarization wave (ATa). APD was measured by four methods. Atrial electrograms were also compared to monophasic action potentials recorded from a dog. ARI computed from atrial action potentials was not very precise, with errors ranging over 30 ms. Determining changes in APD induced by changing [ACh] yielded larger errors. Conversely, ventricular action potentials produced ARIs that very closely correlated with APD, and changes in APD . Positive ATa indicated regions of shortened APD, and islands of ACh release were clearly demarcated by ATa polarity. Experimentally, ARI was able to detect changes in APD, but did not measure APD well. The faster rate of ventricular repolarization produces larger currents that are less susceptible to electrotonic coupling effects, improving correlation with APD. ARI most closely correlated with APD measured as a fixed threshold above rest. Atrial APs produce electrograms that can be used to detect changes in APD. This may be improved by decreasing coupling. The ATa is a robust measure for precisely identifying spatial APD heterogeneities.