Apoptosis is a programmed cell death required for development and maintenance of normal tissue. However, cancer cells often evade apoptosis by inactivating apoptotic genes. Thus, the discovery of apoptotic genes has become important in cancer research. We developed a morphological analysis to quantify apoptosis automatically. Cellular images were acquired by high throughput fluorescence microscopy and normalized. Nuclear and chromatin images were segmented by tophat operations and touching nuclei were further separated by erosion, labeling, and conditional dilation. Shape and intensity features were extracted from nuclear and chromatin images to produce discriminant functions by logistic regression. Nuclear objects were classified as isolated and touching with 91.7% accuracy. Apoptotic and mitotic nuclei were identified with 88.8% accuracy. Automatic apoptotic rates were highly correlated with manual rates (ρ = 0.91, p<0.01). Finally, 20 genes that included seven novel apoptotic genes were found out of 21 genes previously reported to show 3-fold increase of apoptosis.