The general problem of assessing eddy current data from steam generators is complicated due to the presence of random noise and low frequency sinusoidal interference signals. The low frequency sinusoidal interference originates from periodic variations in the material properties or dimensions caused by manufacturing processes, behaviour of the component of the inspection system such as probe wobbling. The estimation of the sinusoidal parameters is obtained by using the cyclic-MUSIC algorithm which is a combination of two well known algorithms, Relaxed and MUSIC algorithms. The removal
process generates waveforms from the estimated parameters and subtracts them from the original data. Application to several data records showed that using all the estimated parameters with two sinusoidal waveforms is the best fit due to the nature of the data. Moreover, the performance assessment was based on power measurements and correlation functions.