Companion diagnostics, tests that purport to classify patients into "responders" and "non-responders" for a specified targeted therapy, demand methods that quantify the actual amount of the corresponding target molecule in tumors from these patients. Various methods are employed depending upon the nature of the target. Many of the candidate therapeutic agents target the abnormal expression of a protein, the detection of which lends itself to an immunohistochemistry (IHC) approach. This review focuses on IHC with formalin-fixed paraffin-embedded (FFPE) tissues for purely pragmatic reasons; first, the morphologic information pertaining to the tumor is of value and should not be discarded as in extraction type assays; second, FFPE tissues are mostly what we have to hand at the time that the diagnostic question is posed. During the four decades of employment of IHC involving the production of a variety of special stains used in the diagnosis or classification of tumors, we have acquired some bad habits, essentially when judging the IHC result via the perception of a "good" stain that "pleases the eye" of the user pathologist, and nothing more. This review takes, as its basic premise, the notion that IHC can be upgraded from its use as a qualitative special staining method to an accurate and reliable quantitative "tissue-based immunologic assay". If accomplished, this enhanced IHC assay would serve accurately to quantify proteins in tissue sections, analogous to the use of the ELISA (enzyme-linked immunosorbent assay) method in the clinical laboratory. The necessary steps for converting IHC to a tissue-based ELISA-like immunoassay of immediate practical use are reviewed with constructive suggestions for steps that can be (must be) taken to achieve the practical reality of quantitative in situ proteomics.