Groups discussing popular topics in online social networks are of great interests recently. In this paper, we measure the dynamics of the online groups discussing incidental popular topics and present method for predicting the dynamic sizes of incidental topic groups. It is found that the dynamic sizes of incidental topic groups follow the law of heavy-tail. Based on the heavy-tailed theory a prediction method is developed for analyzing the dynamics of this type of groups. The models and methods developed in the paper are validated using the actual data from SOHU blog sites, one of the most influential blog sites in China. The experiment results show that the method can predict the dynamic size of incidental topic groups with both short and long time scales.