A new fuzzy complementary filtering algorithm (FCFA) is developed for Unmanned Aerial Vehicle (UAV), which is an accurate, robust and simple algorithm for attitude estimation, and can minimize the errors and reduce computation. The fuzzy complementary filtering algorithm (FCFA) combines the translation attitude estimation from the accelerometer output with the rotation attitude estimation from the gyroscope output. It uses linguistic assessments to adjust the fuzzy weight according to the feedback data from errors and their derivatives to improve the performance and overcome the limitation regarding the rigor and accuracy in traditional complementary filtering algorithm (TCFA). The performance comparisons with TCFA and Kalman filtering algorithm are investigated. Experimental results and fuzzy linguistic analysis are given to shown the good performance of the proposed filtering algorithm.