A method for obtaining distribution functions of relaxation times from impedance data is presented. The method is founded on the principle of maximum entropy in a function closely related to the resistance distribution function. The method is illustrated with a complex real sample.