This study asks (1) are teachers able to identify creative children in their classrooms at better than chance level, (2) does this ability change with grade level, and (3) does creativity itself increase as a function of grade level in school? The Torrance Tests of Creative Thinking--Circles Subtest--was administered to 455 first, second, and third grade children in 18 classrooms. Scores were obtained on the factors of fluency, flexibility, originality, and elaboration. The five high scorers in each classroom were compared to each teacher's judgement of her five most creative students. Creativity was found to increase with grade level, but correct teacher identifications of such children decreased. The authors conclude that although creativity does seem to increase across grades on all factors, teacher identifications were observed to decrease. Teacher ratings would appear to be unreliable criteria for judging students creativity scores on the Circles Subtest of the Torrance Tests of Creative Thinking. (Author)