This study developed mathematical model parameters and optimized the performance of a maize dehusker sheller developed at the Institute for Agricultural Research (IAR), Ahmadu Bello University, Zaria. To facilitate this, a study on some physical properties of ACR, QPM and TZPB maize varieties in relation to moisture content was conducted. Three mathematical models for shelling, grain damage and scatter loss were developed based on dimensional analysis. All model development and validation were enhanced through performance evaluation experiments conducted with speed, feed rate, moisture content and cylinder concave clearance variables each at three levels in a completely randomized block design. The indices of model evaluation used were coefficient of determination, bias, root mean square error, index of agreement, slope and intercept of plots between the predicted values obtained from the developed models and the measured values obtained from experiments. Differential sensitivity coefficients were used to rank variables according to their order of importance in the developed models. SAS package was used for the statistical analysis of variance. Optimization based on genetic algorithm in MATLAB R2008a toolbox was used. Seven (7) functional parameters (input variables) under upper and lower constraints and three (3) performance criteria multi-objective output parameters were considered in the decision process in search of optimal solution. On the study of physical properties, analysis of variance showed that the effects of moisture content and maize grain variety were significant (at 5 %) for all the parameters considered. TZPB variety has the highest mean values of length, width thickness, one thousand grain mass and geometric mean diameter of 1.320 cm, 1.108 cm, 0.492 cm, 375.36 g and 0.896 cm, respectively, at 15 % moisture content, wet basis. The least was the ACR variety with 1.184 cm, 1.027 cm, 0.450 cm, 291.79 g and 0.818 cm respectively, for the parameters. viii The highest mean value for the bulk density (876.40 kg/m3), however, was obtained for ACR variety at 9 % moisture content, wet basis. As regards the dehusker- sheller performance evaluation studies, analysis of variance showed that all the four independent factors have significant effect on all the performance indices for both the model development and validation experiments. The model development mean results for output, shelling and cleaning efficiencies, grain damage and scatter loss were 609.55 kg/h, 98.06 %, and 87.12 %, 4.25 % and 14.08 % respectively. Validation mean results for those indices were respectively 578.26 kg/h, 98.18 % and 94.44 %, 3.97 % and 14.75 %. The different indices of model evaluation namely, bias, root mean square error, index of agreement, square bias and coefficient of determination, between the predicted values and the measured values were: 0.314, 0.194, 0.974, 0.073 and 0.937, respectively for shelling; 0.453, 0.121, 0.907, 0.0144 and 0.901, respectively for grain damage and 1.462, 1.262, 0.629, 1.587 and 0.733, respectively, for grain loss. These show that there was significantly good agreement between the predicted and the measured values. The genetic algorithm optimal values of the input variables were 641.86 kgm-3, 0.617 -1, 0.179 m, 8.70 ms-1, 0.04 m, 10.4 %, and 0.775 m for maize bulk density, feed rate, cylinder drum diameter, cylinder speed, cylinder-concave clearance, grain moisture content (wet basis) and shelling length, respectively. The optimum performance parameters obtained were 99.18 %, 4.23 % and 6.31 % for shelling efficiency, grain damage and grain loss, respectively. Thus parameters for an optimized sheller with maximized shelling and minimized damages and losses have been established.