The objectives of the present study were to investigate the recovery of [14C]allantoin in urine of sheep dosed intravenously and degradation of allantoin by rumen micro-organisms. The recovery of [14C]allantoin in the urine of eight sheep was measured during three periods in two experiments. Individual values of [14C]allantoin recovery varied from 66 to 95 % (mean value 83 (SE 1·6) %). The recovery of [14C]allantoin showed no relation to the level of feed intake. There was some evidence that glomerular filtration rate was an important factor affecting the amount of urinary allantoin recovered in one experiment. Incomplete recovery of plasma [14C]allantoin in the urine indicated losses of plasma [14C]allantoin via non-renal routes. This is supported by the disappearance of 14C from rumen contents incubated in vitro with [14C]allantoin for 48 h (88 %) and the presence of 14C in saliva in vivo from sheep sampled after dosing with [14C]allantoin. However, the amount of 14C activity in the saliva was very low (equivalent to only 1·5 % of the total dose in sheep producing saliva at a rate of 15 litres/d). The proportion of renal and non-renal excretion of purine derivatives was found to be unpredictable both between and within individual animals. The factors responsible for this variability need to be identified, and existing models of excretion of purine derivatives may need to be modified accordingly to improve their accuracy of prediction. A single intravenous injection of [4,5-14C]allantoin provides a simple alternative to infusion methods used to measure the proportion of plasma allantoin excreted in the urine of sheep. Using this method it may be feasible to validate PD excretion models in other ruminant livestock.