Dealing with the assessment of existing masonry structures, there are different sources to acquire the material properties needed for a structural assessment. The most feasible samples but also most destructive are those removed from the building itself. Least destructive tests are those on the components brick and mortar. Many test methods have been developed to retrieve the material properties of the components brick and mortar and in addition, numerical relationships between component characteristics and composite characteristics have been determined, such as homogenization techniques. Disadvantage of these methods is that because of the numerical model involved, an extra model uncertainty has to be taken into account. Tests on bricks and mortar are less time-consuming, less expensive and can be performed in relatively large numbers which enables the probability distribution function of the material properties to be determined. With regard to probabilistic evaluation methods, the probability distribution function of the masonry is looked for. Therefore, the question arises whether the existing or proposed analytical relationships between components and composite are valid for a wide range of values, so it is suitable for statistical processing. This paper will deal with experimental tests and their statistical processing. Main objective is to obtain the probability distribution of the main material properties of masonry needed in a finite element code to evaluate the safety of the structure. Therefore, research is focused on three levels: the components brick and mortar, the composite masonry and the relationship between components and composite. The research focuses on the variability of the composite masonry (distribution type and parameters), the question whether or not the material properties of masonry can be estimated from the components brick and mortar using existing test methods and relationships and the adequacy of the test methods to gather this information on site.