A previous study showed that roughness perception may depend on either temporal or spatial variations in firing rate among cutaneous mechanoreceptive afferents. The present study was designed to distinguish between these hypotheses. Plastic surfaces embossed with patterns of dots designed to produce predictable alterations in temporal and spatial firing rate variation were used as stimuli in psychophysical and neurophysiological experiments. Subjective roughness magnitudes obtained from psychophysical experiments fitted the predictions of the spatial but not the temporal hypothesis. In the neurophysiological experiments, the stimuli were scanned across the receptive fields of cutaneous mechanoreceptive afferents. Firing rate variation in the neural responses was measured using a range of temporal and spatial filters. Temporal variation was not correlated with roughness magnitude. Spatial variation, on a scale of 1-2 mm (one to two receptor spacings), was closely correlated with roughness.