A new multidimensional (3 and 2D) anaerobic digestion model for cylindrical reactor with non-uniform influent concentration distributions was developed to study the way in which mixing intensity affects the efficiency of continuous-flow anaerobic digestion. Batch experiments reported and simulated earlier by Vavilin and Angelidaki (2005) were used to modernize a kinetic scheme and to obtain the corresponding kinetic coefficients. In the new models, hydrolytic microorganisms were included using Contois kinetics for the hydrolysis/acidogenesis degradation of municipal solid waste (MSW). Monod kinetics was applied for description of methanogenesis. Both hydrolytic and methanogenic microorganisms were assumed to be inhibited by high volatile fatty acids (VFA) concentration. According to the new distributed models, the mixing level reduction expressed by increasing dimensionless Peclet number may improve the continuous flow reactor performance at the relatively low influent methanogenic biomass concentration. In the continuously stirred tank reactor (CSTR) there are two steady states with and without methane production at slightly different values of initial methanogenic biomass concentration. In the system, the threshold methanogenic biomass concentration existed because of inhibition by high VFA concentration. High methanogenic biomass concentration is required for efficient anaerobic digestion of MSW in order to avoid possible inhibition due to high VFA build-up. Thus, CSTR configuration might have unstable dynamics at high organic loading as shown in earlier experiments carried out by Stroot et al. (2001). A gradual increase of organic loading during the start up of a completely mixed digester causing an accumulation of methanogenic biomass is a solution to prevent a probable digester failure. According to the distributed models a plug-flow reactor with non-uniform influent concentration distributions where methanogenic and hydrolytic microorganisms are separated has significant methane production and solids removal at the relatively low influent methanogenic biomass concentration. © 2006 Wiley Periodicals, Inc.