The rod-drop experiment has been studied for the purpose of predicting reactor resonance power level. A simplified model, referred to here as the “collective model,” is introduced for experimental analysis of the rod-drop transient response. The mathematical description of this model is formulated by describing the experimentally observed oscillatory response by an overall damping factor and an overall oscillatory frequency. Based on this model, it is found that the overall damping factor is approximately a linear function of the reactor power. Accordingly, we propose an experimental procedure, the method of least-squares approach, which provides an exponential approach to the resonance power level as a function of the number of rod drops. It is shown that the accuracy of measurement in the rod-drop experiment greatly affects this technique for core dynamic analysis. The present results show that for an experiment of negligible experimental error, only two or three rod drops are needed to predict the resonance power level up to an accuracy of 0.2%, while for an experiment of ±5% in error, it requires four to five rod drops to reach an accuracy of 0.8%.