This experiment explored the relationship between neural firing patterns in second somatosensory cortex (SII) and decisions about roughness of tactile gratings. Neural and behavioral data were acquired while monkeys made dichotomous roughness classifications of pairs of gratings that differed in groove width (1.07 vs. 1.90 and 1.42 vs. 2.53 mm). A computer-controlled device delivered the gratings to a single immobilized finger pad. In one set of experiments, three levels of contact force (30, 60, and 90 g) were assigned to these gratings at random. In another set of experiments, three levels of scanning speed (40, 80, and 120 mm/s) were assigned to these gratings at random. Groove width was the intended variable for roughness. Force variation disrupted the monkeys' groove-width (roughness) classifications more than did speed variation. A sample of 32 SII cells showed correlated changes in firing (positive or negative effects of both variables) when groove width and force increased. While these cells were recorded, the monkeys made roughness classification errors, confusing wide groove-width gratings at low force with narrow groove-width gratings at high force. Three-dimensional plots show how some combinations of groove width and force perturbed the monkeys' trial-wise classifications of grating roughness. Psychometric functions show that errors occurred when firing rates failed to distinguish gratings. A possible interpretation is that when asked to classify grating roughness, the monkeys based classifications on the firing rates of a subset of roughness-sensitive cells in SII. Results support human psychophysical data and extend the roughness range of a model of the effects of groove width and force on roughness. One monkey's SII neural sample (21 cells) showed significant correlation between firing rate response functions for groove width and speed (both correlations either positive or negative). Only that monkey showed a statistically significant interaction between groove width and speed on roughness classification performance. This additional finding adds weight to the argument that SII cell firing rates influenced monkey roughness classifications.