A method is developed to generate dependent pseudorandom variates with specified marginal distributions and correlation matrix which makes use of the fractile correlation rather than the usual cross product moment correlation. An example concerning crop yields illustrates how the parameters used by the method can be estimated from historical data.
