In order to evaluate the reliability and reproducibility of the immunohistochemical demonstration of the c-erbB2 oncoprotein and define criteria for positivity assessment in paraffin-embedded material from different tumours, we examined immunohistochemically 79 gastric, 100 colorectal and 20 bladder carcinomas using 5 different c-erbB2 antibodies. To determine criteria for the evaluation of c-erbB2 positivity, the observed immunohistochemical staining was compared with the actual protein prevalence, determined by Western blot analysis of each tumour. The comparison of protein levels obtained from Western blot with corresponding immunohistochemistry revealed that positive staining can only be considered to be specific for c-erbB2 oncoprotein if: (i) there is a definite staining along the tumour cell membrane with no predominant immunoreactivity inside the tumour cell cytoplasm, (ii) the intensity of immunoreactivity is comparable with a positive standard (e.g. in this study breast carcinoma) and (iii) the surrounding normal, non-neoplastic tissue shows no immunoreactivity. The graduation of staining intensities did not correlate to the actual amount of protein detected in Western blots. This study stresses the importance of defined and controlled criteria for evaluation of c-erbB2 oncoprotein positivity in immunohistochemistry.