An alternative for the steady-level flight control on an Unmanned Aerial Vehicle (UAV) is the use of decentralized multiloop PID controllers for each controlled variable. PIDs are linear structured low order controllers which are not so easy to tune in the presence of complex dynamics such as multivariable, nonminimum phase, oscillatory and high-order plants; and the use of conventional design techniques based on linearized models usually does not end in satisfactory results. In this work a design scheme based on iterative feedback tuning (IFT) for the multivariable nonlinear model of the Condor Andino UAV (Andean Condor UAV) is addressed. The method proposes the optimization of a quadratic performance target function using the closed loop response obtained from simulations. In this case the characteristics of the model are not used directly in the controller tuning process, but in simulations and in some other numeric manipulations. The optimization process is made using a modified version of the Levenberg-Marquardt algorithm. The simulation results show that a set of controller parameters can be found such that the target function has a local minimum.