Forty-milligram strips of malignant breast tissue were divided longitudinally into 20-mg microsamples. Soluble protein concentration was determined for one-half, and histologic evaluation for viable carcinoma content was made on its sister half. The correlation coefficient for 88 such comparisons was 0.330. This suggests that estrogen receptor (ER) assays that do not take into account the actual amount of carcinoma present in the sample may not permit reliable stratification of ER data. Using this technique, ER values obtained for 129 breast tumor microsamples were adjusted to reflect carcinoma content. A comparison of ER values before and after such adjustment revealed that the relative status of 73% was not significantly changed. The ER status of 27%, however, was changed sufficiently to be of potential clinical significance.