A computer optimization technique was applied to obtain griseofulvin (GRF)/hydroxypropyl cellulose (HPC) solid dispersions which provide a high dissolution rate and good stability of dissolution properties. The amounts of HPC and methanol were selected as independent variables in the regression analysis (methanol was used for the preparation of the solid dispersions). Concentrations of GRF dissolved in the dissolution medium at each sampling time were selected as dependent variables for deciding the optimum formulation. These dissolution characteristics were predicted quantitatively by using the best combination of independent variables. The physical meaning of the regression equation for each dissolution characteristic was defined by the application of contour graphs. Two hundred mg of GRF, 2 g of HPC and 125 ml of methanol was predicted to be the optimum for the preparation of GRF/HPC solid dispersions. Experimental results with this optimized solid dispersion agreed well with the predictions.