The use of Wireless Sensor Networks is tremendously growing due to their capabilities of computing, communication and their reduced size. Wireless Sensor Networks have a wide spectrum of applications such as smart home, E-parking, E-health� Typically, each sensor consists of a microcontroller, a transceiver, a battery and or any component for saving harvested energy. Moreover, current Wireless Sensor Network platforms like Atmel, CrossBow, and Telos are equipped with a microcontroller with low power consumption. One major challenge is to extend the lifetime of Wireless Sensor Networks. This can be achieved either by power control or energy saving. The idea of this paper is simple and it is related to closed loop power control of XM1000 sensor motes that are equipped with temperature and humidity sensors. It is based on the prediction of the RSSI (Strength of the Received Signal) at the sender side and provides sensors with learning capabilities to adjust power. Experimental results conducted on XM1000 sensor motes that operate under TinyOs illustrate that power control depends on the meteorological conditions (temperature) and has also temporal variations. Controlling power allow also to gain power instead of operating at the maximum power.