The biologic functions mediated by the nuclear protooncogene, c-MYC are correlated to gene dosage. Since automated quantification programs are expensive, time-consuming and not easily available, and since analysis by flow cytometry is difficult in the case of nuclear antigens, we examined the suitability and reproducibility of a semiquantitative in situ evaluation system. This system was based on the percentage of nuclear area staining positively, and comprised the following categories: 0: negative, 1+: single scattered grains of the immunocytochemical staining product, 2+: confluence of grains to patches but less than 50% nuclear area positive, and 3+: greater than 50% positive nuclear area. In addition, sensitivity and specificity of two anti-c-MYC antibodies were investigated. Although both antibodies differed slightly in staining pattern and sensitivity, the four quantification categories were applicable for immunostainings of both antisera and highly reproducible when re-evaluated by the same observer (r = 0.98; p = 0.0001) or a second investigator (rAb155 = 0.98, rAb DCPm = 0.96; p = 0.0001), both reading blindly and independently. Comparing our semiquantitative evaluation categories and results of computer-assisted image analysis, the percentage of positive nuclear area (p less than 0.0001), the median staining intensity (p less than 0.0001), and the product of both (p less than 0.0001) differed significantly in the four evaluation categories. This result still held true after correction for nuclear size, which differed appreciably in various cell types (p less than 0.0001). The product of positive nuclear area, staining intensity and nuclear size (microns 2), which best approximates the absolute amount of c-MYC within a certain cell, was clearly different within the four staining categories (p less than 0.0001) and did not depend on cellular morphology within the staining categories 0 to 2. Also, the immunocytochemical technique proved highly reproducible (median day/day variance 0.65% (0-13); r = 0.995). The practicability of this system for semiquantification was demonstrated by (a) correlation of H score values of immunocytochemical stainings with densitometric scans of Western blots and (b) by the fact that peripheral blood lymphocytes, Phytohemagglutinin stimulated blasts, 13 cases of multiple myeloma and HL-60 cells differed concerning their estimated c-MYC amounts (p = 0.0125). This confirms on the effector molecule level results previously reported from mRNA in situ and Northern blotting analyses. We conclude that a simple and highly reproducible evaluation system can be used for in situ comparison of nuclear oncogene dosage.