Breakdown phenomena in a high power microwave (HPM) system present unique obstacles for the further development of HPM technology. The non-uniformity of a high frequency electric field and the statistics associated with breakdown in general along with the stochastic nature of naturally occurring electron generating mechanisms introduce significant challenges for predicting and preventing breakdown occurrences within a HPM system. An experiment consisting of an S-band multi-megawatt HPM pulse is used for observing an alternating field induced plasma sheath across a dielectric surface. In order to minimize experimental deviations, a continuous UV lamp is used to provide a constant source of initiatory electrons through the process of photoemission. This reduces the waiting time for flashover initiating electrons to appear, however, primarily due to avalanche statistics, variations are still observed. A statistical model that uses an exponential distribution sampling procedure was developed to predict the surface flashover delay times for a variety of conditions. A supporting experiment that uses a continuous UV lamp and a DC electric field is used for measuring low current due to photoemission from the dielectric window. An explanation of the model describing these phenomena is presented along with a comparison of current measurements from the supporting experiment.