Perceptual decisions about ambiguous visual stimuli are influenced by information about both the sensory environment and the value of alternative actions. It is not yet understood where in the neural structures underlying the decision-making process these variables are combined. Introduction of an artificial electrical stimulus into cortical visual area V5/MT of the Rhesus macaque has enabled us to investigate how sensory and value information interact to shape perceptual decisions about structure-from-motion stimuli. Two Rhesus macaques were trained to discriminate the direction of rotation of a structure-from-motion cylinder (duration 2 sec) with an eye movement to one of two choice targets. The stimulus consisted of two transparent surfaces of random dots moving in opposite directions. The two surfaces could be presented at different relative depth, so for some decisions more sensory evidence was available; other trials were ambiguous. Correct responses were rewarded with a fluid reward. Once trained, animals were implanted under general anaesthesia with a headholder and a chamber for access to the cortex (7-8 mg/kg/hr alfaxalone and alfadolone acetate i.v. or 1-3% isoflurane to effect) (see Dodd et al 2001; Read and Cumming 2003 for Methods). During the experiment, V5/MT cortical sites with 200-300 µm of consistent tuning for cylinder’s direction of rotation were identified. Cylinder stimuli were matched to the rotation preference at that cortical site. On 50% of the trials, while the monkey viewed the rotating cylinder, we electrically stimulated the site with a train of 20 µA biphasic pulses. The proportion of reports in one direction as a function of disparity (psychometric functions) were fitted for stimulated and non-stimulated trials with a pair of cumulative Gaussian curves, separated by a horizontal offset or shift. At 27/48 sites, the electrical microstimulation induced a significant shift in the monkey’s perceptual reports in favour of the cylinder tuning preference of the neurons at the stimulated site (Chi-square, p < 0.05). The size of the reward available for a correct report at the end of each trial was varied systematically, such that as the number of consecutive correct choices increased, the size of the available fluid reward increased in two steps up to a maximum. At each stimulation site, trials were split into two conditions, maximal and sub-maximal expected reward size. For trials where the available reward size was maximal, the effect of electrical microstimulation was decreased (Wilcoxon sign-rank test, p < 0.001). This shows that the artificial signal that we inserted directly into sensory cortical area V5/MT was affected differently from stimulus-dependent, non-microstimulated cortical activity as the reward condition changed. The application of a simplified drift-diffusion model of perceptual decision-making to the psychophysical data revealed that in order to explain this result, the effects of expected reward must act on the level of sensory representations prior to the proposed integration of sensory and value evidence into a decision variable in sensorimotor structures. This strongly implies that reward affects representations of sensory evidence in visual cortex, which project to sensorimotor structures.