Models originally compiled to predict nutrient absorption from the digestive tract and metabolized in various tissues could be adapted for CH₄ predictions. Numerous empirical equations and mechanistic models to predict CH₄ emission are available. The Molly cow model is a mechanistic, dynamic model describing digestion and metabolism of dairy cattle with the ability to predict the animal-related factors that affect the environment, including CH₄ emission (Hanigan et al., 2013). The Nordic cow model Karoline is a dynamic, mechanistic model describing digestion and metabolism in dairy cows (Danfær et al., 2006), and it was confirmed by Ramin and Huhtanen (2015) to be a useful tool in predicting CH₄ emissions in cattle. The aim was to evaluate these models for predicting CH₄ emissions in cattle using a data set consisting of 267 treatment means from 55 respiration chambers studies. The data set contained DMI, (14.2 ± 5.82 kg/d); ingredient proportions; dietary contents of CP (156 ± 30.8 g/kg) and NDF (356 ± 105.9 g/kg); BW (531 ± 131.1 kg); and CH₄ (303 ± 118.7 g/d) which covers the range of typical cattle diets. The simulations were conducted using observed DMI, BW and dietary nutrient concentrations and digestion rates. Each treatment mean was simulated and predictions of nutrient digestibility and CH₄ output were collected in a database. The relationships between observed and predicted CH₄ (pCH₄) were assessed by regression analysis. Root mean square error (RMSpE) was calculated as: RMSpE = √ [∑ (Obs – Pred)²/n]. Molly predictions were: CH₄ (g/d) = 0.81 ± 0.018 × pCH₄ (g/d) + 38 ± 6.4 (RMSpE = 54.9 (18.1% of observed mean) CCC = 0.910). The corresponding equation for Karoline was: CH₄ (g/d) = 1.00 ± 0.019 × pCH₄ (g/d) + 5 ± 6.0 (RMSpE = 34.6 (11.4%) CCC = 0.955). Both mean (−27 g/d) and linear bias (−0.19) were significant (P < 0.001) with Molly, but only mean bias (4 g/d) was significant (P = 0.04) with Karoline. Proportions of MSE attributable to mean and linear bias and random error were 23, 24 and 53% for Molly, and 2, 0 and 98% for Karoline, respectively. Based on predictions it can be concluded that both models predicted CH₄ emissions reasonably well in terms of high CCC, but Karoline was more accurate based on smaller RMSE, mean and slope bias.