Abstract: Carbon dioxide (CO 2) is an important raw material of the plant photosynthesis. Increasing CO 2 fertilizer rationally can improve the net photosynthetic rate of plant leaf, and further improve crop yield and quality. To achieve precision management of CO 2 fertilizer in greenhouse, a greenhouse CO 2 fertilizer optimizing control system based on wireless sensor network (WSN) was designed and developed. The whole system includes four monitoring and controlling nodes, an intelligent gateway and a remote management software. The monitoring and controlling node, which connected to sensors and an electromagnet, can real time monitor greenhouse environmental parameters and control the switch of CO 2 source according to the demand of crop. The intelligent gateway can process and transmit the data and commands between nodes and remote management software. It can also storage and display environment parameters locally. Besides, user can control the CO 2 source by gateway. The remote management software, which embeds photosynthetic rate prediction model, can not only process and transmit the data, but also control CO 2 fertilizer remotely. To achieve precision management of CO 2 fertilizer supplement, it was necessary to build an accurate and reliable net photosynthetic rate prediction model. The paper measured environment parameters by the system above mentioned, and obtained single-leaf net photosynthetic rate by LI-6400XT photosynthesis analyzer. Then a photosynthetic rate prediction model based on SVM was established. In order to improve the generality of prediction model, tomatoes in late seedling stage were cultivated in four different fertilizer levels ((700±50)μmol/mol (C1), (1 000±50)μmol/mol (C2), (1 300±50)μmol/mol (C3), ambient about 450 μmol/mol (CK)). The photosynthetic rate prediction model was established by support vector machine (SVM). The environment parameters were used as input variables, and the photosynthetic rate was taken as output variable. The performances of designed system and prediction model were evaluated. The system can work stably and reliably, therefore it can be used to monitor environment information and control the CO 2 fertilizer in solar greenhouse. The prediction results of the model showed that R between predicted and measured data was 0.981 5 and RMSE was 1.092 5 μmol/(m 2 ·s). According to the analysis, it was concluded that the prediction model can be good used as the basis of the quantitative regulation of CO 2 fertilization to tomato plants in greenhouse.