Bearing are machine elements that function as support elements in rotating motion in machine construction. Bearing failure will significantly affect machine performance, therefore it is necessary to monitor the failure growth. This research conducted an experimental study of bearing failure that considered 4 types of experiment variables as factors affecting the natural frequency of bearings, namely bearing type, rotation speed, scratching depth, and scratching location. The scratching shape is deliberately given to the bearing part using laser cutting machine as the failure representation. Scratching was machined on the inner ring raceway, outer ring raceway, or roller element part of the bearing. The Design of Experiment (DOE) used the Taguchi method which aims to get smaller number of trials but could get optimum results. Its experiment was designed with 4 variables and 3 levels and was properly tested on rig testing. The frequency spectrum of machine defect characteristics was measured using a vibration analyzer. The interpreting process and mapping of the frequency characteristics of bearing failure for BPFI, BPFO, BSF, and FTF are using omnitrend software. The resulting research shows that the greater scratching depth causes an increase in the amplitude value of the defect frequency characteristic. The bearing shaft rotation value affects frequency value of the measured bearing defect characteristics.