Solar-power can provide a much needed energy source for wireless sensor networks, which are remotely deployed and limited by the short battery lifetime. A challenge in solar-powered sensor network is that solar power is highly dynamic and volatile: sun light intensity varies significantly with time, with the location of the deployment, the weather conditions, solar panel orientation, and obstacles and shading. Therefore, even if all sensors are equipped with solar panels, individual sensors can receive vastly different amount of solar power and a significant portion of sensors may not receive adequate solar power. This paper addresses such a challenge and introduces a sensor data collection scheme called “Green Diffusion” that automatically adapts sensors data dissemination behavior to the solar power received by the sensors. Performance evaluation shows our scheme significantly reduces sensor battery energy consumption and data latency, even when just a fraction of the nodes receive adequate solar power1.