This research examined children's performance on second‐order false belief tasks as a function of the content area for the belief and the method of assessing understanding. A total of 70 kindergarten and first‐grade children responded to four second‐order stories. On two stories, the task was to judge a belief about a belief, and on two, the task was to judge a belief about an emotion. On one trial within each group, the task was to predict the target's belief, and on one trial, the task was to explain the belief. Older children outperformed younger children on the prediction measure. Differences as a function of content area and method of assessment were limited; when they did occur, performance was generally better with belief than with emotion as the target, and better with prediction than with explanation as the response criterion. Finally, there was no relation between number of siblings and performance.