Pain is a conscious perceptual experience characterized in large part by its aversive qualities and consequent motivation for relief. The Anterior Cingulate Cortex (ACC) is a critical region for the emotional dimension of pain. In both humans and rodents, acute and chronic pain increase ACC neural activity, while ACC lesions or reduced ACC excitability decrease emotional reactivity during pain. However, the ACC receives diverse inputs from a multitude of brain regions, and is engaged and active during a variety of behavioral experiences beside pain. It thus remains unclear what circuit mechanisms in the ACC contribute to shaping pain experience, and how specific those circuits are to nociception. The Periaqueductal Gray (PAG) is a midbrain structure part of descending pain modulation circuits, but without known function in pain affect. Ventrolateral PAG neurons thus control nociceptive inputs in the spinal cord dorsal horn, through the modulation of spinally projecting medullary neurons. Interestingly, the ACC projects to the PAG, but to its dorsal part (dPAG), which is critical for the expression of defensive behaviors and the acquisition of learned responses to aversive stimuli. Together, these studies suggest that the PAG, under the control of the ACC, might have a previously unrecognized function in pain aversion. Here, we tested the hypothesis that excitatory input from the ACC to the dPAG during pain facilitates the affective-motivational dimension of pain. We first examined ACC-dPAG connectivity using histology as well as optogenetics and electrophysiology in PAG slices. We found that ACC axons terminals terminate in dPAG and monosynaptically excite Vglut2+, but not Vgat+ dPAG neurons. Second, we genetically targeted the ACC neurons projecting to the dPAG with viral vectors to express the hM4Di inhibitory DREADD, and exposed the animals to an assortment of pain tests. We found that inhibition of the ACC-dPAG pathway reduced affective-motivational pain behaviors but not sensory-reflexive responses across acute and chronic pain states. Third, we used fiber photometry to record population neural activity in ACC during a variety of pain tasks and found that neurons projecting to PAG are engaged during a broad array of aversive, rather than exclusively painful, experiences, v and with different structures of activity depending on the task context. Finally we optogenetically inhibited ACC-dPAG neurons during social observation of pain and observed alterations in non-somatically induced pain experience. Collectively, these results uncover the direct contribution of ACC-dPAG neural activity to the aversive experience of pain, and the necessity of this pathway to generate aversive behavioral responses. These results are most supportive of a theory of the ACC contributing a feature of general value appraisal, rather than specifically encoding the unpleasant quality of noxious stimuli.