Abstract: Efficient, accurate, real-time and stable link quality estimation is essential to guarantee the upper layer protocol performance in wireless sensor network. This study expounds a new LQI-based (Based on Link Quality Indicator) link quality estimation mechanism for wireless sensor networks, which takes the lost packets and error packets into account. Based on real platform, we carry out a large number of experiments. After being analyzed the collected data, a new link evaluation model between LQI and PRR (Packet Received Ratio) is established. Experiment results have validated the correctness of the LQI-PRR model and the results also show that compared with traditional assessment methods, the mechanism proposed in this study improves the accuracy of the assessment and is more useful. At last, we analyze BER (Bit Error Ratio) and draw a relationship between LQI-PER (Packet Error Ratio), LQI-BER in order to get a further understands of the link quality.