We measured the ability of humans to discriminate the positions of spherical objects passively contacting the fingerpad. The discrimination threshold averaged 0.55 mm for a moderately curved sphere (radius 5.80 mm) and decreased to 0.38 mm for a more curved sphere (radius 1.92 mm); since the receptor density is about 1 per mm2, these values are substantially smaller than those predicted by the sampling theorem (referred to as hyperacuity). To elucidate the underlying neural mechanisms, responses to the same spheres and random sequences of stimuli were recorded from single Merkel afferents (SAIs) and Meissner afferents (RAs) in anesthetized monkeys. For multiple applications of identical stimuli, coefficients of variation of responses were around 3%. Profiles of responses across the SAI population were “hill-shaped.” A change in position of the stimulus on the skin resulted in a matching shift of the profile, evident over the whole profile for the more curved sphere but ony at the skirts for the less curved sphere. The shift in response profiles, relative to the standard deviations, increased as the change in position increased, and was more reliable for the more curved sphere. Responses were measured over four time frames: 0.2, 0.3, 0.5, and 1.0 sec. Although responses increased with an increase in integration time, so, too, did their standard deviations, so that signal-to-noise ratios or the resolution in the SAI population was bout the same at 0.2 sec as at 1.0 sec. Only half the RAs responded; responses were small, but signalled reliable information about the position of the stimulus.