The paper is concerned with the justification of Monte Carlo simulation schemes for the solution of nonlinear kinetic equations via time averaging. Some results obtained recently about systematic errors inherent to these methods are reviewed. Sufficient criteria for their validity are derived, and tools reducing the effect of particle correlations are presented.