Clinical follow-up and normal plain radiology are common protocols for detecting implant loosening in Total Hip Replacement (THR), which cannot detect it at very early stages. There are more accurate radiological techniques (arthrography, radio-stereometric analysis) which are not widely employed due to their cost and invasiveness. The shift detection of natural frequencies by vibration analysis has been also used as a more sensitive alternative. However, interpreting the data is not easy in clinical environments. An experiment with three stem models with a very different design in artificial femurs was performed in this work. Strain gauges were employed as the dynamic sensor, and free decay bending and bending + torsion tests were performed for two extreme situations including completely loose and well-fixed stem. The strain gauges signal was processed in time-domain and frequency domain, and resonant shifting and increases in dumping ratio were analyzed. Resonant shifting was not easy to recognize, especially for some models. However, dumping ratio showed very significant changes in all models and did not need transforming signals into frequency-domain for its computing. The findings indicated that dumping ratio is a highly suitable parameter and easier to compute in searching for implant loosening.