Fermentation in the rumen, measured by a steady state in vivo method, was compared to in sacco estimates obtained by different methods of calculation. These methods included a first-order model calculated over 24 h, a first-order model calculated over 48 h, a first-order model incorporating lag time, a moment-generating function of a gamma distribution and the calculation of mean retention time (MRT) by a discrete approximation to the mean of a continuous random variable. A comparison of in vivo and in sacco fermentation showed, in general, good agreement between the two methods. Care should, however, be taken when selecting a method to calculate in sacco estimates, as some methods induce large differences between the in vivo and in sacco estimates. The MRT method, based on the discrete approximation of a continuous random variable, is recommended since estimates obtained in this manner are explicable in terms of rumen dynamics.