THE SIXTH-GRADE STUDENTS OBTAINING THE 80 HIGHEST MEAN SCORES ON THE METROPOLITAN ACHIEVEMENT TEST FORM D WERE RANDOMLY ASSIGNED TO GROUPS OF EASY OR DIFFICULT TASKS WITHIN EACH OF FIVE DECIBEL LEVEL GROUPS. THE LEARNING TASKS WERE TAKEN FROM THE MEANINGFULNESS OF ALL POSSIBLE CVC TRIGRAMS (ARCHER, 1960). THE APPARATUS INCLUDED A TAPE RECORDING OF A CONSTANT WHITE NOISE AND THE EARPHONES TO ELIMINATE ALL NOISE FOR CERTAIN GROUPS. A TWO-WAY FIXED EFFECTS ANALYSIS OF VARIANCE WITH REPLICATIONS WAS CARRIED OUT. ON BOTH THE EASY AND DIFFICULT LEARNING TASKS, PERFORMANCE WAS CURVILINEAR, INCREASING FROM THE 0 TO THE 55 DECIBEL LEVEL GROUP, AND DECREASING FROM THE 55 TO THE 85 DECIBEL LEVEL GROUP. SIGNIFICANT DIFFERENCES WERE FOUND--(1) BETWEEN THE 0, 20, AND 55 DECIBEL LEVEL EASY TASK GROUPS, (2) BETWEEN THE 55, 70, AND 85 DECIBEL LEVEL DIFFICULT TASK GROUPS, AND (3) BETWEEN PERFORMANCE ON THE DIFFICULT AND EASY LEARNING TASKS WAS HIGHEST AT THE 55 DECIBEL LEVEL FOR A CONSTANT WHITE NOISE. STUDENTS LEARN BEST WITH A MODERATE AMOUNT OF NOISE, AS COMPARED TO NO NOISE, OR EXCESSIVE NOISE. THIS PAPER WAS GIVEN AT THE 1968 ANNUAL MEETING OF THE AMERICAN EDUCATIONAL RESEARCH ASSOCIATION (CHICAGO, FEBRUARY 7-10). (PS)