The aim of eddy current inversion is to reconstruct an unknown flaw from probe signals measured as a function of probe position and frequency or time. In seeking the solution of an inverse problem, one can take advantage of an ability to solve forward problems by using predictions of probe signals calculated with a tentative estimate of the flaw. A well known inversion strategy adopted here for eddy current inversion, is that of varying the flaw estimate iteratively until the disparity between predicted and observed probe signals is minimized. Then the model flaw which gives the optimum predictions of the observations should approximate the actual flaw.