Time-order error, a psychophysical phenomenon in which the duration in between successive stimuli alters perception, has been studied for decades by neuroscientists and psychologists. To date, however, the locus of these effects is unknown. We use intracortical microstimulation of somatosensory cortex in three humans with spinal cord injury as a tool to bypass initial stages of processing and restrict the possible locations that signals could be modified. Using a 2-interval forced choice amplitude discrimination paradigm, we first assessed the extent to which order effects are observed. Comparing trials where the standard stimulus was in the first or second interval, we found that systematic biases are exhibited, typically causing the intensity of the second stimulus to be overestimated The degree of this overestimation for individual electrodes was dependent on the perceptual sensitivity to changes in stimulus amplitude. To investigate the role of memory on this phenomenon, we implemented a 2-interval magnitude estimation task in which participants were instructed to ignore the first stimulus and again found that the perceptual intensity of the second stimulus tended to be enhanced by the first in a manner that depended on the amplitude and duration of the first stimulus. Finally, we repeated both paradigms while varying the inter-stimulus interval to examine the timescale over which these effects occur and found that longer inter-stimulus intervals reduced the effect size. These results show that direct activation of primary somatosensory cortex is sufficient to induce time-order errors.