The conventional theory of electrophoretic processes is based on a model in which the migration process is determined by a biased diffusion process which predicts that the peak shape should be Gaussian. This implies that the peak position is proportional to time and the width (measured as the full width at half-maximum) is proportional to (time)(1/2). Recently developed instrumentation allows a more critical examination of both the assumptions and conclusions derivable from theory by intermittent scanning of the migration path. Data toward this end were collected on the kinetic behavior of phycoerythrin in an agarose gel. These data indicate significant deviations from a Gaussian shape and from results derivable from a model based on diffusion. A number of parameters that could be measured rather simply were compared with the predictions of both a diffusion-based theory and a non-Markovian theory, showing that the latter produced results in better accord with experiment. A practical implication of this finding is that the resolution of two peaks, after an initial transient period, need not increase with migration time.