In scatterometry, a merit function including a regularization parameter is used in an iterative process to find values for the scattering properties of the measured target. An optimal value for the regularization parameter is obtained for each measurement target and in each iteration of the iterative process. Various methods can be used to find the value for the regularization parameter, including the Discrepancy Principle, the chi-squared method and novel modifications of the Discrepancy Principle and the chi-squared method including a merit function.