Subjects made magnitude estimates of line length in a short-term memory (STM) task in which the standard and comparison lines were presented for 1 sec each, separated by a 2-sec interval (Experiment 1). Response time (RT) increased systematically with judged magnitude. These findings are consistent with previous results for perceptual magnitude estimation of length (Hartley, 1977, 1981) and show that the effect takes place in STM as well. To explain this effect, Hartley has proposed that subjects repeatedly lay off a mental image of the standard line along the line to be judged. Similar results were found in Experiment 2, in which subjects made magnitude estimates of the loudness of a 1-kHz tone preceded by a standard. Since the relation between RT and judged magnitude appears to hold for the nonspatial continuum of loudness, Hartley’s visual imagery model may be too specific to account for the RT effect in the general case. Two alternative explanations are considered: (1) a sensory image model based on Hartley’s proposal and (2) an abstract analog model with a nonsensory representation.