The practicability of quality assurance in immunohistochemistry and its integration into the diagnostic process were both tested in this Germany-wide interlaboratory trial. One hundred seventy-two pathologists received one hematoxylin and eosin and five unstained slides from five cases; all cases were selected by a panel because immunohistochemistry was required for their final diagnosis. Participants rendered a morphologic diagnosis and then substantiated it immunohistochemically. Stained slides and evaluation sheets were reviewed by the panel, and the diagnostic process was analyzed in individual steps: morphologic diagnosis, selection of antibodies, staining quality, interpretation of stained slides, conclusions, and final diagnosis. Diagnosis-independent immunohistochemical performance was tested using a multisample tissue block (30 samples) that was stained and evaluated for six common antigens. For individual cases, corresponding to their difficulty, 21-89% of the final diagnoses (altogether 57% from 828 diagnoses) were correct. In a statistical analysis, the tentative diagnosis, the interpretation of stains and conclusions drawn from immunohistochemistry, were independent factors in reaching the diagnosis. Sensitivity to detect estrogen receptors on the multisample tissue block was only 48%. However, 24% of the stains were interpreted as falsely negative. The low staining sensitivity was not correlated to the number of correct diagnoses. The major problem of applying immunohistochemistry in surgical pathology appears to be its integration into the diagnostic process and not the staining quality. Both future quality control projects and training will have to regard these integrative requirements. Multisample tissue blocks provide a promising tool to standardize quantitative immunohistochemical parameters, such as receptor or proliferation scores.