In this dissertation we present an automatic flaw detection system for heat exchanger tubes in steam generators. The system utilizes two well-known techniques, wavelets and fuzzy logic, to automatically detect the flaws in tubing data. The analysis of eddy current inspection data is a difficult task which requires intensive labor by experienced human analysts. To aid the analysts, an accurate and consistent automatic data analysis package was developed. The software development is divided into three parts: data preprocessing, wavelet analysis, and a fuzzy inference system. The data preprocessing procedure is used to set up a signal analysis standard for different data and also to remove the variations due to lift-off and other geometrical effects. The wavelet technique is used to reduce noise and identify possible flaw indications. Due to multiresolution and the unique time-frequency localization properties of the wavelet transform, the flaw signals have specific characteristics in the wavelet domain. We fully utilize those characteristics to distinguish flaw indications from noise. To further evaluate the flaw candidates and reduce false calls, we invoked fuzzy logic to discriminate between true positives and false positives. A template matching technique and fuzzy inference system were developed. The template matching technique uses signals from artificial flaws as templates to match with possible flaw signals and execute a normalized complex crosscorrelation. Through this process, we obtain both phase and shape information which are placed into a fuzzy inference system for final decision making. A rigorous test of the system using actual inspection data was undertaken. Results from tests indicate that the new techniques show a great deal of promise for automatic flaw detection. Investigating the novel techniques and integrating them into a system are the major contribution of this work.