Maintaining accurate bug priorities is an important part of software maintenance. Assigning the appropriate priority is not always easy, and these priorities can change as bugs are investigated. To improve this process, we have applied naive Bayes and logistic regression to identify bugs with the incorrect priority. These algorithms were trained on 3,000 bugs from the Mozilla Firefox bug repository submitted over the last two years. Using the bug change history we created snapshots of each bug when they were one week old, and converted the summary and comments into a weighted bag-of-words vector. Approximately 2% of bugs had the incorrect priority since they were re-prioritized at some point in the future. We achieved 5% precision and 50% recall in predicting these incorrectly prioritized bugs.