Methods of vibration diagnostics have been widely used throughout many years for early detection of the failures of rolling element bearings. This method is based on the comparison of the results of vibration measurements (e.g. certain scalar values over a pre-determined frequency range, or time signals, or frequency spectra, etc.) against rules, and threshold values recommended by international standards. Depending on the results of the comparison the state of the bearing is considered as good, as acceptable or as not acceptable. This is a simple, but not always efficient method to detect faults, especially in early stage of bearing damage. This is the reason why other methods, including detailed vibration signature analysis, etc. have been developed and used all over the world. These methods of vibration analysis usually need the identification of the so called bearing defect frequencies, which can be calculated using mathematical formulas. The goal of our research was to verify the accuracy of calculation of these bearing defect frequencies, using artificial faults made on both the inner and outer rings of the bearing. For the tests 6 pieces of deep groove ball bearings (of type 6205), mounted in a test rig, driven by a variable speed electromotor rig have been used. Vibration signals have been recorded in three directions (axial, horizontal, and vertical), at three different rotational speeds (500 rpm, 1000 rpm and 1500 rpm), using Envelope Acceleration, HFD (High Frequency Detection) and SEE (Spectral Emitted Energy) methods, and 17 different measurement setups. Altogether 918 measurements have been carried out. The results were evaluated on time signal records, vibration spectra, including waterfall diagrams and Palograms. As a conclusion, it was found that the accuracy of bearing defect frequencies is acceptable, and the best method for an early warning of potential bearing failure is the so called Envelope Acceleration signal.