This paper presents the results of system identification (system ID) of a small fixed-wing low-cost unmanned aerial vehicle (UAV). The system ID estimates the aerodynamic coefficients of the linear longitudinal equations of motion using batch least squares enhanced with an Error Filtering Online Learning (EFOL) scheme. EFOL estimates the state derivatives using filtering without taking the derivative, making the least squares more robust. Thus, this method of system ID only requires flight data of the states and not the state derivatives. This is advantageous due to the limitations of low-cost sensors including GPS-aided INS systems. The derivation and benefits of least squares with EFOL are briefly described here in. The main contribution of this paper is the identification of the aerodynamic force and moment coefficients for the linear longitudinal model of the UAV. The error between the flight data and the model response are also presented. The identified model shows good correlation to the training data. The model predicts well the short period flight characteristics of the fixed-wing UAV but does not fully capture the long period dynamics. Solutions to this issue are presented and discussed.