The efficacies of four methods, used for the prophylaxis of cobalt deficiency in sheep as measured by the elevation of liver and serum vitamin B12 levels, were compared in marginally deficient sheep over 14 weeks. The methods used were weekly drenches of either cobalt sulphate or cobalt chelate (EDTA) three-weekly injections of hydroxocobalamin, and ruminal cobalt pellets. On the basis of elevated liver and serum vitamin B12 levels, chelated cobalt was shown to be available to rumen microflora for the synthesis of vitamin B12. However, at no stage were liver and serum vitamin B12 levels of sheep receiving the chelate significantly different from those receiving the same amount of cobalt as the sulphate. After five, three-weekly injections of hydroxocabalamin liver vitamin B12 levels were significantly higher (p<0.01) than for the other treatments, with the exception of cobalt sulphate. Cobalt pellets led to an initial rapid and significant rise in serum vitamin B12 when compared with the other treatments. However, at four weeks there was no significant difference between treatment groups for serum vitamin B12. Fourteen weeks after the administration of cobalt pellets, serum and liver.vitamin B12 levels in this group were not significantly different from those of untreated sheep. At this time, three out of 12 sheep had lost their pellets.