The reproducibility of measurements of oestrogen-receptor activity has been examined in multiple specimens from a rabbit uterus, a rat mammary tumour and human breast tumours. The relationship between receptor concentration and tumour histology has also been investigated in 11 large primary tumours. In the animal tissues, receptor measurements were relatively reproducible (coefficient of variance: wet wt. basis 16-17%, protein basis 16-21%) but in human breast tumours receptor activity varied considerably (c.v.: wet wt. basis, 22-125%; protein basis, 28-72%). In addition to these variations in receptor activity within tumours, there was a difference between tumours, as demonstrated by an analysis of variance (P less than 0.01). In the 11 primary breast cancers selected for study, the level of receptor activity was related to menopausal status and the tumour content of the specimen. We conclude that the receptor activity detected varies within a tumour and depends upon the tumour content of the biopsy specimen. Predictions based on precise quantitation of receptor concentrations may therefore necessitate replicate tumour sampling and correction for the fraction of non-tumour tissue in each sample.