The predicted outcome value of an action influences both decision, should the subject execute the action or not, and vigor, how quickly to act. Previous studies have shown that the activity in the striatum is correlated with the temporally discounted value of the upcoming reward. We wish to characterize how the coding of discounted value contributes to both the decision to act and the vigor with which the action occurs. We recorded neural activity in the caudate nucleus by using a reward-size-and-delay task (Minamimoto et al., 2009). The monkey obtained a reward if it correctly released a bar after red dot changed to green. A visual cue presented throughout each trial indicated how long the reward would be delayed after the trial was completed (0, 3.3, or 6.9 s) and how large the reward would be (1 or 3 drops of reward). Error rate and reaction time are influenced by both the combination of the forthcoming reward size, the delay and satiation. The error rate is well explained by E= (1+kD)/aRe^-λS, where D is delay to reward, R is reward size, S is satiation level and k, a and λ are fitting parameters. We recorded 99 phasically active neurons in the caudate nucleus of 2 monkeys. 64 neurons increased their activity after the cue; 15 showed activity correlated with the discounted value (DV) estimated from subject's decision, DV = R/(1+kD). The DV coding started just after the monkey was informed by the cue about the size and delay to reward, and continued until execution of the action. The cue response did not change between the 1st and 2nd half of the recording (120~ trials), suggesting that activity was insensitive to satiation. Bilateral inactivation of the caudate nucleus by local injection of muscimol (dose; 4μg/μl, volume; 3μl, n=5) altered the error rate pattern, but, did not affect reaction times. The combination of the physiology and muscimol results suggest that the discounted value signal in the caudate is used to formulate a decision about whether to act or not.