We determined the effect of the core sample dimensions on the measured values of the Schmidt rebound hardness and proposed a way to correct it to obtain unaffected value. The effect was previously investigated on cubes, where a clear dependence of the rebound hardness on the cube edge was proved. For cylindrical bodies, however, the problem is more complex. In order to determine and verify our model, a set of cylindrical samples was made from greywacke. After applying the correction, an improvement in the mean rebound hardness of the whole set was achieved from 80 to 98.8 % of the unaffected mean rebound value.