Rates of methane (CH4) production and the sources of carbon (C) for its synthesis were studied in four mature ewes when dosed with a CH 4-mitigating ionophore ICI-111075, or monensin, or when untreated. The sheep were given 700 g/day of chaffed lucerne hay in equal portions every hour, before and during experiments in which 14C-labelled NaHCO3-, acetate, propionate, lactate and formate were infused intraruminally over 12 h and the specific radioactivity of C (SR) in each of these substrates was determined. During these infusions, the SR of material in the primary pool (the tracer infusion site) and in secondary metabolites of this material (secondary pools) approached asymptotic or 'plateau' values. The rate of infusion of 14CH4 (kBq/day) divided by the plateau SR value (kBq/g C) gave estimates of the rate of irreversible loss of CH4 (g C/day). These calculations indicated that CH4 production rate was reduced by 72% when sheep were dosed with ICI-111075 and by 58% when dosed with monensin. With monensin, the reduction in CH4 production was not associated with hydrogen (H2) accumulation in the rumen headspace gases whereas with ICI-111075, the decrease in CH4 production was associated with marked H2 accumulation in the headspace gases. When plateau SR were attained during the tracer infusions, the percentage ratio, (SR in any secondary pool: SR in the primary pool) gave an estimate of the fraction of C in that secondary pool derived from material of the primary pool. Calculated in this way, the percentage of CH4-C derived from rumen fluid carbon dioxide (CO 2) averaged 59% in untreated sheep, and when sheep were dosed with ICI-111075, the corresponding percentage averaged 12%. These findings indicate there are sources of C for rumen CH4 synthesis other than rumen fluid CO2. However, there was no evidence that C from acetate, propionate, lactate or formate in rumen fluid were direct sources of the unidentified CH4-C. One plausible explanation for these findings is that CH 4 is synthesised within naturally occurring microbial biofilms (attached to feed particles or the rumen wall) from CO2 that is produced locally by fermentation of unlabelled substrates within the biofilms. It is postulated that such pools of CO2 would be kinetically distinct and, during the infusion of 14C-labelled substrates, would exhibit a lower SR than the CO2 in the surrounding rumen fluid.