Four vitamin B 12 assays were compared using blood sera from Friesian cows on winter diets or grazing. In herd 1, ten animals were blood-sampled three times at monthly intervals and the vitamin B 12 concentration of the sera determined by the Poteriochromonas malhamensis and Lactobacillus delbrueckii assays. At all three sampling dates the results produced by the P. malhamensis assay were significantly greater than those produced by the L. delbrueckii assay. Cows in herd 2 were divided into two groups, each of 30 animals. One group was given a soluble glass bolus releasing cobalt and the other was unsupplemented. Milk yields were recorded throughout the experimental period and herbage samples were taken at intervals for cobalt determinations. Blood samples were taken at intervals from January to September 1983 and the vitamin B 12 concentration of the sera determined by four different assay methods: the P. malhamensis and L. delbrueckii microbiological assays and the Becton Dickinson and RIA Products 'No-boil' radioassays. The last of these failed to detect vitamin B 12 in any sample. There was a significant difference between the results obtained by the three other assays, with the Becton Dickinson radioassay consistently producing the lowest result and the P. malhamensis assay the highest result of the three. There was no significant effect of cobalt supplementation on milk yield.The vitamin B 12 requirement of ruminant animals is assumed to be satisfied when adequate cobalt is supplied in the diet. The currently recommended level of cobalt for dairy cows is 0-11 mg/kg dry matter (DM) (Agricultural Research Council, 1980), which is an increase of 10% on the Agricultural Research Council (1965) recommendation. A recent report (Agricultural Development and Advisory Service et al. 1983) indicates that a level of 0-2 mg/kg DM may be preferable.The adequacy of dietary cobalt is often assessed by the vitamin B 12 status of the animal. Several assays are used to measure the vitamin B 12 content of animal tissues, but the normal range quoted varies with the assay. This variation can be related to the fact that there are vitamin B 12 analogues as well as true vitamin B 12 molecules present in these tissues (Rickard & Elliot, 1978Elliot, 1979), yet only true vitamin B 12 is metabolically active for higher animals. The measured values of vitamin B 12 appear to differ depending on the degree of specificity of each assay for 17 DAR 54