The earth's climate system has strong natural and periodic or quasiperiodic oscillation because of adjustments by its internal dynamics especially ocean dynamics. In addition, the climate system possesses long-term trend as a result of external forcing such as anthropic greenhouse gas emission. Eliminating the influence of natural and periodic oscillations in estimating the long-term trends is a key concern in climate change studies. To remove or reduce the influence of natural and periodic oscillations on trend estimation, we propose a method-the running trend method, based on an idealized mathematical model. The fundamental theory and its feasibility are expounded. We tested the method using the idealized theoretical model, and the result shows that the method can realized an estimation very close to the true trend. The running trend method is applied to estimate the trend of Indonesian Throughflow transport during 1985~2010, and the results suggest that the true trend is nearly double of the rough trend estimation, which is of much importance in understanding the Indonesian Throughflow and Indo-Pacific climate system. The running trend method is of clear universality, and it can be widely used in the study of the earth's climate change and trend estimations of any variables.