Software to track sample lanes automatically in four-color, fluorescence-based, electrophoretic gel images has been developed for application in large-scale DNA sequencing projects. Lanes and lane boundaries are tracked by analyzing a first difference approximation to the gradient of a vertically integrated and processed "brightness" profile. Initially lanes are located in a region of the gel image selected for good horizontal lane spacing and signal strength. The software uses models of expected lane and interlane spacing and lateral lane behavior to maintain accurate tracking on imperfect gels. In areas where intensity-based tracking is difficult, interpixel column correlation is also used to locate and define lane features. Summary statistics and compressed-in-time images are generated for user evaluation of tracking performance. The software developed has been tested successfully on gel images with degradations including significant horizontal lane motion (curving) and image artifacts, and is now in full-scale use in our sequencing projects.