a b s t r a c tThe sulphur hexafluoride (SF 6 ) tracer technique enables determination of enteric methane emissions from large numbers of individual ruminant animals. The objective of this research was to identify and correct substantial errors within the SF 6 technique. Six experiments were undertaken using respiration chamber, laboratory or SF 6 techniques. Experiment 1 used respiration chambers to demonstrate that the daily pattern of methane emissions from dairy cows was related to their pattern of feed intake. In contrast, the daily emission of SF 6 from these cows was constant and independent of the pattern of methane emission. This finding supports the contention that in order to accurately determine daily methane emissions using the SF 6 technique, it is necessary that gases are collected continuously at a constant rate for 24 h. Since development of the SF 6 technique in 1993, it has been propounded that capillary-tube flow restrictors achieved a constant rate of sample collection into evacuated gas collection canisters. Laboratory experiments 2, 3, 4 and 5 demonstrated that, when capillary-tube flow restrictors are used, the rate of sample collection declined and caused a bias of up to 15.6% in calculated methane emissions. This error was caused by an interaction between the declining sample collection rate and the pattern of an animal's methane emission over 24 h. In contrast, orifice plate flow restrictors maintained a constant sample collection rate at canister pressures <0.31 atm and thereby minimised the decline in sample collection rate. Experiment 5 also demonstrated that sample collection using orifice plate flow restrictors, combined with initial (<0.03 atm) and final (<0.49 atm) canister pressures, substantially reduced measurement error. Accuracy of the modified SF 6 technique, incorporating orifice plate flow restrictors for 24 h sample collection, was validated in Experiment 6. The mean (S.D.) methane yield (g CH 4 /kg DMI) of eight cows did not differ (P = 0.135) when determined using the modified SF 6 technique 22.3 (1.44) or chambers 21.9 (1.65). In addition, the between-animal coefficient of variation for methane yield determined using the modified SF 6 technique (6.5%) was similar to that determined using