Identification experiments using 13 tone‐pulse stimuli (1000 Hz, 500 ms) spaced by equal decibel increments from 42 to 90 dB SPL were conducted under three conditions: all intensities equally likely (denoted Q), the middle intensity presented on roughly 1/3 of the trials (M), and each of the extreme intensities presented on roughly 1/5 of the trials (E). Experiments were performed both with and without feedback, and results were analyzed in terms of sensitivity and bias. Similar results were obtained with and without feedback: listeners generally shifted biases in the M and E conditions relative to the Q condition to increase percent‐correct scores, but there were only minor changes in sensitivity from condition to condition. However, sensitivities were generally higher and biases more optimal when feedback was available. The results on sensitivity are consistent with those associated with payoff variation [Lippmannetal., J. Acoust. Soc. Am.59, 129–134 (1976)], but are inconsistent with those reported for presentation probability variation in tone frequency identification [Cuddyetal., J. Exp. Psychol.100, 218–220 (1973)]. In an auxilliary experiment concerned with tone frequency identification, results consistent with those obtained for intensity, but inconsistent with those reported by Cuddyetal., were obtained.