In long-term time-lapse imaging of living cells, where recording takes several minutes or longer, a drift of focus may be significant. Focus-drift is due to the slippage in the microscope focus mechanism and/or the thermal gradients in the microscope. Software and hardware solutions may be introduced to correct for the focus-drift. Some autofocus techniques measure position of the specimen by sensing the light or sound reflected from a well-defined surface, such as the microscope slide. An autofocusing approach, where a focus measure is computed for images acquired at different objective positions is less appropriate in confocal microscopy, since more than one section is in focus. To correct for the focal-drift in long-term time-lapse confocal imaging, we acquired an image stack of the specimen periodically. The software calculated Pearson's correlation coefficient between each image in the z-stack and the reference image in the stack, which was selected at the beginning of the experiment. The maximal correlation coefficient of pixel intensities was taken to identify the image, which corresponded to the focal plane of the reference image. To test our approach, we used confocal images of living rat lactotroph cells, which discharged preloaded green fluorescent probe from a single secretory granule. Simultaneously, an extracellularly applied FM 4-64 red fluorescent probe loaded the discharging vesicle and the plasma membrane. We show that our approach is appropriate to correct for focal-drift in long term time-lapse imaging and analysis of living cells.