A model-free PID controller tuning approach is presented in this paper. The method can handle different stability and performance indicators in the frequency domain. The phase margin, gain margin, crossover frequency and more advanced indicators which are the infinity-norm of the sensitivity functions can be considered as well for the design. The actual values of the design parameters are measured directly on the system thanks to closed-loop experiments. A frequency criterion is then defined as the weighted sum of squared errors between the measured and desired values of the design parameters. The minimization is done iteratively using the Gauss-Newton algorithm. The approach presented does not require any parametric model of the plant and can be applied to a wide range of industrial applications. Simulation examples show the rapid convergence of the algorithm and the effectiveness of the method for PID controller tuning.