Aerosol filtration in fibrous filters is one of the principal methods of accurate removal of particulate matter from a stream of gas. The classical theory of depth filtration of aerosol particles in fibrous structures is based on the assumption of existing single fibre efficiency, which may be used to recalculate the overall efficiency of entire filter. Using “classical theory” of filtration one may introduce some errors, leading finally to a discrepancy between theory and experiment. There are several reasons for inappropriate estimation of the single fibre efficiency: i) neglecting of shortrange interactions, ii) separation of inertial and Brownian effects, ii) perfect adhesion of particles to the fibre, iv) assumption of perfect mixing of aerosol particles in the gas stream, v) assumption of negligible effect of the presence of neighbouring fibres and vi) assumption of perpendicular orientation of homogenous fibres in the filtration structure. Generally speaking, “classical theory” of filtration was used for characterization of the steady - state filtration process (filtration in a clean filter, at the beginning of the process) without deeper investigation of the influence of the nternal structure of the filter on its performance. The aim of this review is to outline and discuss the progress of deep-bed filtration modelling from the use of simple empirical correlations to advanced techniques of Computational Fluid Dynamics and Digital Fluid Dynamics.