Using the method of based reasoning data fusion, Dempster-Shafer evidence reasoning applies to class of oval fruits recognition experiments. Through multiple sensors to observe and sample the fruits, the characteristic information is extracted, which can determine the basic incredible assigned of various sensors for the unknown fruits. And then Dempster combination rule was used to fuse the data further and improve reliability; lastly, to recognize the type of fruit effectively. This experiment shows the D-S evidence reasoning has a strong ability to deal with uncertain information, the result proved the correctness and effectiveness.