The common pathway of any apoptotic cascade leads to the activation of the so-called execution caspases, particularly caspase-3 (CSP3). The question of whether immunohistochemical (IHC) detection of activated CSP3 might be useful for routine purposes still needs to be clarified. We analyzed apoptoses in gastrointestinal graft-versus-host disease (GvHD) and cytomegalovirus (CMV) colitis using a commercially available polyclonal antibody against activated human CSP3. In GvHD samples obtained from the colon, apoptoses detected by CSP3 varied between 11 and 43/40 crypts, and in esophageal specimens between 21 and 40/1.5 mm, squamous epithelium basal length. This count was correlated with the apoptotic count assessed on hematoxylin- and eosin (H&E)-stained slides. A perfect concordance for those counting between 30 and 40 apoptoses/40 crypts or 1.5 mm squamous epithelium basal length was detected, whereas cases with low apoptotic counts on the H&E stained slides showed a 2 to 3 fold greater number of stained nuclei as assessed by CSP3 staining. In CMV-colitis, although the number of exploding crypt cells was 8-13/40 crypts, only 1-2 nuclei/40 crypts and almost all cells with typical nuclear inclusions stained positively. The presence of CMV can be easily detected on H&E- or IHC-stained slides, while masked GvHD by an overlying CMV-colitis might remain unrecognized. Staining for CSP3 may be helpful in distinguishing these two conditions, as apoptotic count would be excessive in GvHD.