Feed intake (FI), live weight (LW), and ADG were recorded over 31 d in ninety-six 12-month-old ewes (progeny of 4 sires) given ad libitum access to chaffed lucerne/cereal hay. Methane (CH) and CO emissions of each ewe were measured for 40 to 60 min in portable accumulation chambers (PAC) and in respiration chambers (RC) over 22 h. Testing in RC increased the variability of FI on the test day and depressed the amount eaten from an average of 1,384 to 1,062 g/d; FI depression increased by 0.63 ± 0.24 percentage points for every kilogram of additional LW. Repeatabilities of PAC measurements were 0.76 (CH) and 0.81 (CO). After adjusting for LW and ADG, repeatabilities were 0.47 (PAC CH) and 0.43 (PAC CO). Daily FI measurements had similar repeatability (0.76 before and 0.42 after adjustment for LW and ADG). The PAC measurements were highly correlated with mean 31-d FI ( = 0.81 for both CH and CO). After adjustment for LW and ADG, PAC measurements were moderately correlated with residual feed intake (RFI; = 0.37 for CH, 0.31 for CO). The CH:CO ratio was also significantly correlated with mean 31-d FI ( = 0.52). After most of the ewes had given birth and raised lambs, repeat PAC measurements were available for 91 of the ewes at 2 years of age (with ad libitum access to the same feed). Correlations with the 2012 PAC measurements were 0.64 (CH) and 0.75 (CO). After adjusting 2014 PAC measurements for LW, correlations with RFI in 2012 were 0.34 (CH) and 0.33 (CO), with a clear relationship between sire means for RFI in 2012 and PAC CH adjusted for LW in 2014. These results suggest that PAC tests under similar feeding conditions are repeatable over an extended time period and can provide useful information on FI and feed efficiency as well as methane emissions. Analyses of RC measurements might need to consider FI depression.