We investigate how accurate parents are at predicting their children's meal preferences and what cues best describe parents' predictions. In Study 1, 30 parents predicted their children's school lunch choices from actual school menus. Parents' prediction accuracy matched the stability of children's meal choices (assessed in a 4-month retest), implying that accuracy was as high as can be expected. Parents appeared to make their predictions by using specific knowledge about their child's likes and by projecting their own preferences. In Study 2, we asked 58 parents to predict their children's preferences for 30 randomly drawn school meals, and compared them to the children's actual preferences. Again, parents showed high prediction accuracy and predicted the lunches their children liked correctly more often than the disliked ones. Overall, parents' accuracy in predicting their children's food preferences was as good as or better than found in previous preference prediction studies that used less ecologically relevant task domains.