This paper presents a comparison of the performance of four different sampling methods used in stochastic simulation. The methods are crude Monte Carlo sampling, median LHS, Halton sequence and a new method based on gaussian quadrature. A model of a batch distillation column is used to compare the accuracy and the convergence properties of the various methods. The results show that the;quadrature method is superior the others in this example. A convergence rate up to 1250 times faster than of MC sampling is found.