A methodology is developed to identify and flag significant trends related to the safety and availability of U.S. commercial nuclear power plants. The development is intended to aid in reducing likelihood of human errors. To assure that the methodology can be easily adapted to various types of classification schemes of operation data, a data bank classified by the Transient Analysis Classification and Evaluation (TRACE) scheme is selected for the methodology;The significance criteria for human-initiated events affecting the systems and for events caused by human deficiencies were developed;Clustering analysis was used to verify the learning trend in multi-dimensional histograms. A computer code is developed based on the K-Means algorithm and applied to find the learning period in which error rates are monotonously decreasing with plant age;The Freeman-Tukey (F-T) deviates are used to select generic problems identified by a large positive value (here approximately over 2.0) for the deviate. The identified generic problems are: decision errors which are highly associated with reactor startup operations in the learning period of PWR plants (PWRs), response errors which are highly associated with Secondary Non-Nuclear Systems (SNS) in PWRs, and significant errors affecting systems and which are caused by response action are highly associated with startup reactor mode in BWRs. Those are corresponding to inconsistencies in the pattern of associated data. The program P3F for the analysis of multiway frequency tables in Biomedical Computer Programs may provide incorrect estimates for the expected frequencies if structural zeros, where the frequency is constrained to be zero, occur. Therefore, to get reasonable results, efforts to eliminate structural zero entries in multiway tables should be made. This may be accomplished by redefining the multiway tables or by using a more general computer program for the multiway tables.