The characteristics of antigen- and antibody-coated enzymeimmunoassay (EIA) formats to measure oestrone sulphate (OS) were studied using a murine monoclonal antibody as the primary antibody. In an antigen-coated format the most sensitive EIA (9 fmol/well) was achieved using 6-ketoestrone-6-O-carboxymethyloxime (OCMO) coupled to bovine serum albumin (BSA), as the coating antigen, and horseradish peroxidase (HRP), as the enzyme label. In an antibody-coated format, comparable sensitivity could be achieved using HRP conjugated to either OCMO, oestrone-3-glucuronide (OG) or oestrone-3-hemisuccinate (OHS) as the steroid 'tracer'. In both the antigen- and antibody-coated formats replacing HRP with alkaline phosphatase (AP) markedly aggravated the assay sensitivity. The antigen-coated EIA format was used to measure OS concentrations in cow's milk directly without an initial defatting step, and a progressive increase in OS concentrations in milk as pregnancy progressed was observed. Median OS concentrations rose from 1.1 nmol/l at days 70-99 of pregnancy (n = 44) to 3.2 nmol/l at days 140-160 (n = 92). Oestrone sulphate concentrations in milk from non-pregnant cows (n = 51) ranged from non-detectable to 1.3 nmol/l with a median value of 0.4 nmol/l. Only 5% of cows 120 or more days pregnant had milk OS concentrations within the range of values found in milk from non-pregnant cows. Accurate discrimination of non-pregnant and pregnant cows can thus be achieved on the basis of OS concentrations in milk samples taken at least 120 days after mating/insemination. This EIA for OS may have a role in the dairy industry as an alternative non-invasive means of determining pregnancy status in cows.