Development of effective methane (CH) mitigation strategies for grazing sheep requires accurate prediction tools. The present study aimed to identify key parameters influencing enteric CH emissions and develop prediction equations for enteric CH emissions from sheep offered fresh grass. The data used were collected from 82 sheep offered fresh perennial ryegrass () as sole diets in 6 metabolism experiments (data from non-grass-only diets were not used). Sheep were from breeds of Highlander, Texel, Scottish Blackface, and Swaledale at the age of 5 to 18 mo and weighing from 24.5 to 62.7 kg. Grass was harvested daily from 6 swards on contrasting harvest dates (May to December). Before the commencement of each study, the experimental sward was harvested at a residual height of 4 cm and allowed to grow for 2 to 4 wk. The feeding trials commenced when the grass sward was suitable to zero grazing (average grass height = 15 cm), thus offering grass of a quality similar to what grazing animals would receive under routine grazing management. Sheep were housed in individual pens for 14 d and then moved to individual calorimeter chambers for 4 d. Feed intake, fecal and urine outputs, and CH emissions were measured during the final 4 d. Data were analyzed using the REML procedure to develop prediction equations for CH emissions. Linear and multiple prediction equations were developed using BW, DMI, GE intake (GEI), and grass chemical concentrations (DM, OM, water-soluble carbohydrates [WSC], NDF, ADF, nitrogen [N], GE, DE, and ME) as explanatory variables. The mean CH production was 21.1 g/kg DMI or 0.062 MJ/MJ GEI. Dry matter intake and GEI were much more accurate predictors for CH emissions than BW ( < 0.001, = 0.86 and = 0.87 vs. = 0.09, respectively). Adding grass DE and ME concentrations and grass nutrient concentrations (e.g., OM, N, GE, NDF, and WSC) to the relationships between DMI or GEI and CH emissions improved prediction accuracy with values increased to 0.93. Models based on farm-level data, for example, BW and grass nutrient (i.e., DM, GE, OM, and N) concentrations, were also developed and performed satisfactorily ( < 0.001, = 0.63). These models can contribute to improve prediction accuracy for enteric CH emissions from sheep grazing on ryegrass pasture.