The liver consists of numerous repeating, randomly oriented, more or less cylindrical units, the lobules. Although enzyme-histochemical or microbiochemical assays accurately reflect zonal differences in lobular enzyme content, their results cannot be directly compared to biochemical assays. This is because section-based assays typically sample along a linear portocentral column of cells, even though periportal regions contribute substantially more to hepatic volume than pericentral regions. We have developed a time-efficient approach that depends on image analysis to determine the prevalence of hepatocytes (pixels) with a defined cellular concentration of a particular gene product (absorbance), and that generates a graph with the average absorbance per hepatocyte on the ordinate and the percentage of hepatocytes with absorbances in each of a predetermined range of absorbances incrementally summed on the abscissa. The direction of the gradient is read directly from the section. The gradient is a graphical representation of the two-dimensional distribution pattern of the gene product between the portal tracts and the central veins. The total surface area underneath the resulting graph represents the integrated absorbance and is equivalent to the outcome of a biochemical assay. The typical linear portocentral gradient can be derived from that representing the two-dimensional distribution if we assume that liver lobules are uniformly cylindrical or prismatic. The analysis, therefore, yields a quantitative description of the relation between the enzymatic phenotype of hepatocytes and their position on a normalized portocentral radius. We have used the procedure to compare portocentral gradients of different enzymes in the same liver and of the same enzyme in different livers. In addition, bipolar portocentral gradients of the same enzyme in the same liver were analyzed.