Microwave links are widely used for wireless data exchange, particularly between base stations of mobile phone networks. Because of the range of frequencies used for data transmission, the link signal is attenuated when rainfall occurs along the link path. This attenuation can be related to the path-averaged rain rate. A critical issue in this procedure is the ability to separate the attenuation due to rainfall from the attenuation occurring during dry periods. This letter presents a technique to separate dry and rainy periods and to estimate a time-varying attenuation baseline using path-integrated attenuation measurements from operational telecommunication microwave links. Dry and rainy periods are separated by analyzing the local variability of the link signal. The attenuation baseline is estimated in real time through the classification into dry and rainy periods. The method is applied to four different links and ten different rain events. Measurements from a nearby C-band weather radar are used to evaluate the performance of the algorithm. The new method performs well, identifying about 92% of all rainy periods, 86% of all dry periods, and 93% of the total rain amount on average.