This report presents the rules of tests modifying the electronic adjudication (eAdjudication) business rules to attempt to reduce the rate of false alarms and to incorporate potential applications of natural language processing (NLP). eAdjudication is a tool used by the Federal Government to automatically grant favorable eligibility determinations to clean cases. Because the rules are quite conservative, eAdjudication frequently generates false alarms, incorrectly transferring clean cases to human adjudication rather than granting a favorable determination. This report presents the results of tests to reduce this false alarm rate without increasing the risk of incorrect favorable determinations. In addition, preliminary NLP results are presented that reproduce certain business rules using the unstructured text contained within the background investigation. Results of the business rule testing suggest two potential modifications to reduce false alarms without increasing security risk. First, modifying the rule that selects only cases with a case seriousness code of “G” to include cases with case seriousness codes of “R” and “A” as well. Second, deactivating the rule that checks the results of the question on the Standard Form 86 about Selective Service registration. Making these changes reduced false alarms by 8.4%. In addition, NLP results show that criminal history can be predicted with some reliability using the unstructured text from the Report of Investigation. Future research should explore the possibility of including NLP or other avenues for increased complexity of business rules in order to further reduce the rate of false alarms.