Bug reports are widely used in several research areas such as bug prediction, bug triaging, and etc. The performance of these studies relies on the information from bug reports. Previous study showed that a significant number of bug reports are actually misclassified between bugs and nonbugs. However, classifying bug reports is a time-consuming task. In the previous study, researchers spent 90 days to classify manually more than 7,000 bug reports. To tackle this problem, we propose automatic bug report classification techniques. We apply topic modeling to the corpora of preprocessed bug reports of three open-source software projects with decision tree, naive Bayes classifier, and logistic regression. The performance in classification, measured in F-measure score, varies between 0.66-0.76, 0.65-0.77, and 0.71-0.82 for HTTPClient, Jackrabbit, and Lucene project respectively.