Respiration chambers are considered the reference method for quantifying the daily CH production rate (MPR) and CO production rate (CPR) of cattle; however, they are expensive, labor intensive, cannot be used in the production environment, and can be used to assess only a limited number of animals. Alternative methods are now available, including those that provide multiple short-term measures of CH and CO, such as the GreenFeed Emission Monitoring (GEM) system. This study was conducted to provide information for optimizing test procedures for estimating MPR and CPR of cattle from multiple short-term CH and CO records. Data on 495 Angus steers on a 70-d ad libitum feedlot diet with 46,657 CH and CO records and on 121 Angus heifers on a 15-d ad libitum roughage diet with 7,927 CH and CO records were used. Mean (SD) age and BW were 554 d (SD 92) and 506 kg (SD 73), respectively, for the steers and 372 d (SD 28) and 348 kg (SD 37), respectively, for the heifers. The 2 data sets were analyzed separately but using the same procedures to examine the reduction in variance as more records are added and to evaluate the level of precision with 2 vs. 3 min as the minimum GEM visit duration for a valid record. The moving averages procedure as well as the repeated measures procedure were used to calculate variances for both CH and CO, starting with 5 records and progressively increasing to a maximum of 80 records. For both CH and CO and in both data sets, there was a sharp reduction in the variances obtained by both procedures as more records were added. However, there was no substantial reduction in the variance after 30 records had been added. Inclusion of records with a minimum of 2-min GEM visit duration resulted in reduction in precision relative to a minimum of 3 min, as indicated by significantly ( < 0.05) more heterogeneous variances for all cases except CH4 in steers. In addition, more records were required to achieve the same level of precision relative to data with minimum GEM visit durations of 3 min. For example, in the steers, 72% reduction in initial variance was achieved with 30 records for both CH and CO when minimum GEM visit duration was 3 min, relative to 45 records when data with a minimum visit duration of 2 min were included. It is concluded from this study that when using records of multiple short-term breath measures of CH or CO for the computation of an animal's MPR or CPR, a minimum of 30 records, each record obtained from a minimum GEM visit duration of 3 min, are required.