Abstract: In order to improve the reliability of wireless data transmission of a greenhouse environment data acquisition system, this paper studied the reliability between WSN nodes in several cases in a greenhouse based on an experimental method. When the communication quality was affected by distance, obstacles, and high humidity, the relationship between packet loss rate and RSSI (Received Signal Strength Indication) was analyzed, and then RSSI was used to evaluate the reliability of communication. In the experiment, the distance between nodes had been set from 5 m to 40 m at intervals of 5 m, and the humidity varied from 35% RH to 80% RH at intervals of 15% RH. The obstacles included tomato plants, hanging strawberry, greenhouse facilities such as shade net, heat insulation nets, and exhaust fans. The experiment results showed that the packet loss rate increased when the distance between nodes extended or the obstacles existed while it was not affected by the humidity in the greenhouse. In these circumstances, when the transmitting power of the node had been set as 0, 4, 8, 12, or 19 dBm, both the RSSI and packet loss rate changed so that the relationship between the RSSI and packet loss rate could be studied in coordinates. The results included two situations. The first was when there was no obstacle. In those cases, we found that: 1) with the increase of the RSSI of the receiving node, the packet loss rate changed with a certain trend to decrease; 2) when the RSSI value was greater than ?58 dBm, the packet loss rate was almost zero. The second was when there were different kinds of obstacles. In those cases, we found that: 1) packet loss rate decreased when the RSSI got smaller. 2) when packet loss rate was about 1%, for different obstacles, the RSSI values varied from ?58 dBm to ?50 dBm. 3) for the same RSSI, it was the smallest when there was no obstacle. Based on this study, an adaptive transmitting power control algorithm for WSN nodes was proposed in which RSSI was used to evaluate communication quality, and transmitting power was enhanced to improve the reliability of communication. This algorithm included two steps. First, it assumed that there was no obstacle, and the transmitting node estimated the RSSI of the receiving node with its own RSSI. If the estimated value was lower than ?55 dBm, the transmitting node would increase its power while if the estimated value was far more than ?55 dBm, the node would decrease its power to save energy. Second, the algorithm compared the actual packet loss rate to the reference input, so the algorithm could be corrected, depending on the error of the two. The algorithm had been tested in the greenhouse when the communication distance was 5 m, 20 m, and 40 m, and it also had been tested when obstacles existed such as tomato plants, suspended strawberry, and exhaust fans. In the worst situation, the packet loss rate was 2.2%. In addition, a contrast experiment was conducted to show that the algorithm could set the transmitting power at a low level when the communication quality was fine. For example, when the distance was 5 m and no obstacle existed, the transmitting power was set to 0 dBm, which is the smallest one of all of the available transmit values. The research provided an approach to enhance the communication quality of WSN in greenhouse under unfavorable conditions that made progress on the application of WSN to realize wireless data collection in a greenhouse.