Ca antigen levels in serum samples from three groups of patients were assayed. From a survey of 173 patients with various malignancies, elevated levels were found most consistently in patients with metastatic breast cancer. Spearman rank correlation values of Ca and CEA values on individual serum samples, 0.3009, (n = 194), or individual and serial samples, 0.2406, (n = 264) from a total 194 patients with metastatic breast cancer showed that correlation between Ca and CEA values was poor. For a group of 20 patients within the 194, from whom fortnightly serial samples were available, serum levels for 10/20, measured retrospectively, corroborated clinical observations on the course of their disease, although only 4/20 showed marked elevations during active disease. No correlations between expression of the tumour marker and histological type of the primary tumour, age of the patient, site of recurrence nor aberrant elevation in response to cytotoxic drug could be found to explain the non-correspondence of marker behaviour and disease status in the remaining 10 patients. The indications from this small study are that serial Ca antigen serum measurement could be misleading in 50% of patients with metastatic breast cancer, and that the assay is unsuitable for follow-up of patients with breast cancer.