Both short-term and long-term biofiltration experiments were undertaken with a biofilter inoculated with a defined microbial consortium and treating an alkylbenzene mixture. The results obtained with such a biofilter in short-term experiments were very similar to those obtained with a biofilter inoculated with a non-defined mixed culture, in terms of maximum elimination capacities (70-72 g m(-3) h(-1)) and the corresponding removal efficiencies (>95%). However, in long-term experiments, a better performance was reached, with a maximum elimination capacity of 120 g m(-3) h(-1), corresponding to a removal efficiency >99% after 2 years of operation. Inoculation proved to be useful for shortening the start-up period. In the long term, it appeared that biomass distribution was not homogenous along the biofilter, which in some cases resulted in a bad fit between simple model equations and experimental data.