The desk calculator has been promoted as invaluable to the teaching of mathematics, though this claim has not been supported by controlled experimental studies. Using empirical methods, this study was designed to test calculators with two groups of ninth grade students and one group of fifth grade students in three differing instructional settings. Experimental and control classes were designated for each group, and, in addition, one ninth grade group contained a class to control for the "Hawthorne effect." An attempt was also made to control for differences in teacher performance. The findings were equivocal concerning the effect of calculators on the students' performance and on their self-confidence and attitude toward mathematics. Teacher enthusiasm for use of the calculators bore no relation to the machines' effect on student performance. Rather, this enthusiasm appeared to be a function of the calculators' effect on student-teacher interaction and their potential as toys to provide classroom diversion. The lower the average IQ level of the group, the higher was teacher enthusiasm. At worst, the machines took up space and cramped daily operations. At best, they gave a feeling of increased productivity to students not previously capable of it, thereby easing classroom behavior problems. (GO/MT)