Comparative analysis (CA) of dynamical systems is an important problem in qualitative reasoning. CA techniques predict differences in the behavior of two systems as a consequence of differences in the initial conditions or structural differences. A disadvantage of these techniques is the imprecision of the possible answers due to their qualitative nature. This report presents SQCA, an implemented technique for the semi-quantitative comparative analysis of dynamical systems. SQCA is both able to deal with incompletely specified models and make precise predictions by exploiting numerical information in the formof interval bounds on variable values and envelope functions around monotonic relations. The technique has a solid mathematical foundation which facilitates proofs of correctness and convergence properties.