Many feeding trials have been conducted to quantify enteric methane (CH4) production in ruminants. Although a relationship between diet composition, rumen fermentation and CH4 production is generally accepted, the efforts to quantify this relationship within the same experiment remain scarce. In the present study, a data set was compiled from the results of three intensive respiration chamber trials with lactating rumen and intestinal fistulated Holstein cows, including measurements of rumen and intestinal digestion, rumen fermentation parameters and CH4 production. Two approaches were used to calculate CH4 from observations: (1) a rumen organic matter (OM) balance was derived from OM intake and duodenal organic matter flow (DOM) distinguishing various nutrients and (2) a rumen carbon balance was derived from carbon intake and duodenal carbon flow (DCARB). Duodenal flow was corrected for endogenous matter, and contribution of fermentation in the large intestine was accounted for. Hydrogen (H2) arising from fermentation was calculated using the fermentation pattern measured in rumen fluid. CH4 was calculated from H2 production corrected for H2 use with biohydrogenation of fatty acids. The DOM model overestimated CH4/kg dry matter intake (DMI) by 6.1% (R
2=0.36) and the DCARB model underestimated CH4/kg DMI by 0.4% (R
2=0.43). A stepwise regression of the difference between measured and calculated daily CH4 production was conducted to examine explanations for the deviance. Dietary carbohydrate composition and rumen carbohydrate digestion were the main sources of inaccuracies for both models. Furthermore, differences were related to rumen ammonia concentration with the DOM model and to rumen pH and dietary fat with the DCARB model. Adding these parameters to the models and performing a multiple regression against observed daily CH4 production resulted in R
2 of 0.66 and 0.72 for DOM and DCARB models, respectively. The diurnal pattern of CH4 production followed that of rumen volatile fatty acid (VFA) concentration and the CH4 to CO2 production ratio, but was inverse to rumen pH and the rumen hydrogen balance calculated from 4×(acetate+butyrate)/2×(propionate+valerate). In conclusion, the amount of feed fermented was the most important factor determining variations in CH4 production between animals, diets and during the day. Interactions between feed components, VFA absorption rates and variation between animals seemed to be factors that were complicating the accurate prediction of CH4. Using a ruminal carbon balance appeared to predict CH4 production just as well as calculations based on rumen digestion of individual nutrients.