The paper focuses on optimizing the elastic polymer-cement protection coating composition using material model. The material model consists of mathematical relations between composition and technical features demanded for protection, which are: flexibility, waterproofness and vapor permeability. The optimization that enabled obtaining the desirable values of features was proceeded by using statistical experimental design. Taking data from tests and choosing material variables, a material model based on quadratic functions was determined and in consequence three top rank compositions – each for a given feature – have been formulated. Evaluation of compositions was done using the desirability function approach. The individual desirability functions for the particular feature and the combined overall desirability function were treated as additional properties of the coating. Also, after defining the relation between desirability and composition, the overall desirability was enclosed in the material model as its generalization. High values of overall desirability confirmed the usefulness of compositions. The investigation proved that material models based on statistical methods are efficient tools for designing polymer-cement protective coatings.