Evaluating mechanical properties of existing constructions represents a challenging task, especially when dealing with historical masonry buildings, on which diagnostic investigations should be as less invasive as possible. In the last decades, the possibility of testing masonry cores to derive compressive and shear properties has been investigated, showing promising results, supporting the diffusion of this testing technique. However, there are still no specific guidelines that define standard procedures for the application of these testing methodologies. In more detail, with reference to compression tests on cores, they can be performed considering several core geometries and bond patterns and adopting different setups. To be tested in compression, cores should be capped with a high-strength material, with the objective of simulating the confinement given by the masonry on-site and providing regular surfaces on which the compressive load can be applied. In literature, different characteristics of the cap, in terms of material and geometry, have been considered, but the study of the correlation between cap properties and core test results is still a matter of debate. For this reason, in this work, compression tests on masonry cores are carried out with the aim of evaluating the influence of the cap material and geometry on the test results, so to assess the reliability of the compression test on cores in estimating the compressive strength of masonry. Double-wythe cores, having a diameter of 100 mm and two different bond patterns (with a horizontal mortar joint or with a horizontal and a vertical mortar joint), are tested, considering a steel or a high-strength mortar cap and different cap geometries, in terms of length and height. The core test results are compared with the results of standard compression tests to evaluate the reliability of the test methodology and identify the parameters which could mostly influence the test outcomes.