Serum vitamin B12 from 6 sheep fed a diet containing 0.06 mg cobalt per kg dry matter (DM) was analysed by a radioisotope dilution (RID) technique. The results were compared with values obtained by a microbiological method (Lactobacillus leichmannii - ATCC 7830) specific for "true" vitamin B12. Serum vitamin B12 concentrations were four to six times greater during the 16 weeks experimental period with the RID method as compared to the microbiological technique. During the course of the experiment, values obtained by the RID method remained superior to the threshold value (200 pg B 12/ml) of the microbiological method. This investigation indicates that the RID method is not suitable for the rapid diagnosis of vitamin B12 deficiency in subclinically cobalt-deficient sheep.