The existence of a wide reinforced concrete (r.c.) building stock, which is reaching the end of its service life, has focused the attention on its vulnerability assessment. The first step of this analysis is the estimation of the concrete compressive strength. In this field, many codes allow to supplement the classical destructive tests with nondestructive ones, because of their versality and of the possibility of reducing the time spent for tests. It is worth noting that the spread of these tests is strictly connected to the accuracy of the conversion models, which correlate the nondestructive measurements to the concrete strength. The present paper deals with this issue by examining the results of the experimental investigations, which made use of destructive and ultrasonic pulse velocity tests, on three r.c. buildings in Bari (Italy). The data are analyzed and rearranged to define several testing conditions, for each one of which a conversion model is calibrated. Moreover, the variability of the parameters of the identified models is discussed with respect to the number of data considered in the calibration process, to the chosen model, and to the accuracy of the assessed strength.