Property prediction for a given polymer structure using group contribution models require that the structure can be fully represented by groups with well-defined contributions for that particular property. Frequently this cannot be accomplished. To overcome this limitation a group contribution+ approach is developed here, for polymers. With this approach the missing groups are created and their contributions are predicted using zero- and first-order connectivity indices. To minimize the time spent on computational aspects of model development, grid technology is used in this work. Property prediction methods for linear polymers are presented for the amorphous and crystalline volumes, glass transition temperature, solubility parameter and refractive index.