The segmentation of three-dimensional microscopic images of car- diac tissues provides important parameters for characterizing cardiac diseases and modeling of tissue function. Segmenting these images is, however, chal- lenging. Currently only time-consuming manual approaches have been devel- oped for this purpose. Here, we introduce an efficient approach for the semi-automatic segmentation (SAS) of cardiomyocytes and the extracellular space in image stacks obtained from confocal microscopy. The approach is based on a morphological watershed algorithm and iterative creation of wa- tershed seed points on a distance map. Results of SAS were consistent with re- sults from manual segmentation (Dice similarity coefficient: 90.8±2.6%). Cell volume was 4.6±6.5% higher in SAS cells, which mainly resulted from cell branches and membrane protrusions neglected by manual segmentation. We suggest that the novel approach constitutes an important tool for characterizing normal and diseased cardiac tissues. Furthermore, the approach is capable of providing crucial parameters for modeling of tissue structure and function.