Many terror attacks occur at the beginning of electoral terms. We present a game theoretical model with incomplete information to account for this empirical pattern. Both terrorists and governments can be of weak or strong types. We find that the risk of terror attacks is highest at the beginning of electoral terms, because striking early allows the terrorists to collect valuable information about the government's type, and also because terrorists know that even initially weak governments sometimes retaliate to show toughness closer to an upcoming election. The model's predictions are consistent with anecdotal evidence.