The performance of atmospheric pressure photoionization (APPI) and ESI for CZE was compared using a set of seven drugs (basic amines, quaternary amines and steroids) and four different BGEs. The influence of volatile and nonvolatile BGEs of acidic and neutral pH on the MS responses of test compounds was evaluated by infusion of test solutions into the respective ion sources, and by actual CZE-MS experiments. The infusion experiments indicate that sodium phosphate buffers cause ionization suppression in ESI-MS, although for the amines the suppression was modest (25-60% signal reduction). By contrast, APPI-MS responses were not affected by nonvolatile BGEs. With phosphate buffers, ESI-MS responses for the basic amines were still a factor 3-13 higher than the APPI-MS signals, whereas the steroids yielded similar responses in ESI-MS and APPI-MS. The quaternary amines could readily be detected in ESI-MS, but detection in APPI-MS required specific interface conditions. Using typical CZE-APPI-MS settings, quaternary amines remained undetected. Remarkably, the S/Ns observed in CZE-ESI-MS for the test compounds, were generally similar when using volatile and nonvolatile BGEs. For basic compounds, the S/Ns obtained in CZE-ESI-MS were a factor 2-5 higher than in CZE-APPI-MS, whereas steroids yielded equal S/Ns in both methods. Overall, it is concluded that when using relatively low BGE concentrations, the sensitivity of ESI-MS detection in CZE is more favorable than APPI-MS detection, even when nonvolatile BGEs are employed.