SUMMARY: Brown's original method [1955] for the chemical determination of oestrone, oestradiol-17β and oestriol in human urine gives inaccurate results when applied to the urine from patients with breast cancer being treated with cortisone after oophorectomy-adrenalectomy or hypophysectomy. This is demonstrated by comparing chemical results with biological assays of the activity in the phenolic fraction of the urine. There is no evidence of any material in the phenolic fraction interfering with the biological assay. Results with the modified chemical method [Brown, Bulbrook & Greenwood, 1957a], which obviates the chemical interference caused by cortisone metabolites in the urine, may still disagree with the biological results. There are two main reasons for this. The modified chemical method is still subject to error if the patients are taking laxatives, though the biological assay is unaffected. Furthermore, most of the patients are excreting amounts of oestrogen at or below the limits of accuracy of the chemical method. In the absence of interference by laxatives, biological and chemical results usually agree when the calculated errors of the chemical method at these low levels of oestrogen excretion are taken into account. Biologically detectable oestrogen was present in the phenolic fraction of the urine from eleven out of fifteen oophorectomized-adrenalectomized or hypophysectomized women.