The purpose of this study was to compare four measures of difficulty of mental multiplication items: percentage of pupillary dilation, latency of solution, number of correct responses, and judgment of item difficulty. Sixteen multiplication problems, classified into four levels of difficulty, were presented visually to 13 Ss. The Ss verbalized their solutions to the problems. Analyses of variance and correlation coefficients were computed. It was concluded that all four measures of difficulty were useful but that judgment of difficulty and latency of solution were better measures of item difficulty than were the other two. A discussion is presented on various aspects of the pupillary response curve, as well as suggestions for more appropriate measuring techniques.