Confidence Intervals for NONMEM Fitted Curves

Does anyone out there have some generalized code for calculating the confidence limits of the predicted concentrations from a population analysis. This is certainly possible given estimates of the theta's, eta's, sigma and a variance-covariance matrix, but it would be nice to have a generalized method within the NM-TRAN code (or even an option within NONMEM itself) to calculate these values "on the fly" so that they could be tabulated and plotted. Any suggestions?... Brian Sadler