One approach to conveying tactile feedback from sensorized neural prostheses is to characterize the neural signals that would normally be produced in an intact limb and reproduce them through electrical stimulation of the residual peripheral nerves. Toward this end, we have developed an integrate-and-fire model that predicts with millisecond accuracy the timing of responses of the mechanoreceptive afferents that innervate the glabrous skin of the hand. Individual afferents produce highly repeatable and stereotyped responses to a given stimulus. However, responses differ considerably across afferents, even across afferents of a given type. In the present study, we wish to assess the extent to which this within-type variability shapes the signal conveyed by the hand to the brain. Specifically, we wish to determine the extent to which a single canonical model can be used to mimic the responses of a population of afferents during a set of activities of daily living. We find that the spiking responses simulated using the canonical model does not match, in their fine temporal structure, those simulated using individually fit models. However, population firing rates simulated using a canonical model match those simulated using individual models. Our results suggest that afferent heterogeneity is important if the read-out of the response of afferent populations is sensitive to the precise temporal structure of the population response. To the extent that precise spike timing (at a resolution of milliseconds) is not essential, a canonical model can be used to simulate the responses of populations of afferents.