The long-range objective of the work reported here is to provide a theoretical basis for the prediction of the probability of flaw detection in eddy current nondestructive evaluation (NDE). As demonstrated in a previous communication [1], much of the labor involved in probability of detection analyses can be transferred to a computer if one has available a reliable algorithm for the prediction of flaw signals as a function of flaw size and shape, probe geometry, and the other parameters defining an eddy current inspection. Because there is no simplifying symmetry in the interaction of a general eddy current field with a flaw of arbitrary shape and position, the model used for flaw signal predictions must be three dimensional, and capable of predicting the probe impedance change for a flaw at an arbitrary position in the field of an eddy current probe. The immediate objective of the present work is to develop such a three-dimensional model.