We study a Monte Carlo algorithm for computing marginal and stationary densities of stochastic models with the Markov property, establishing global asymptotic normality and O(n^(1/2)) convergence. Asymptotic normality is used to derive error bounds in terms of the distribution of the norm...