Self-administered electrical brain-stimulation reward is one of the most powerful reinforcers known, rivalled only by the most intensely addicting drugs (e.g. cocaine). In humans, such stimulation produces intense pleasure or euphoria. The brain systems subserving this reward apparently consist of synaptically interconnected neurons associated with the medial forebrain bundle (MFB). ‘First-stage’ neurons run caudally within the MFB, and synapse in the ventral tegmental area on ‘second-stage’ dopaminergic neurons running rostrally within the MFB, which are preferentially activated by habit-forming drugs, and which synapse in the nucleus accumbens on ‘third-stage’ endogenous opioid peptide neurons. Many other types of neurons synapse onto this reward circuit to regulate hedonic tone. Also, this reward circuit is strongly implicated in the pleasures produced by natural rewards (e.g. food, sex). It is widely assumed that crawing is mediated by these same circuits. Some theories posit that craving results from neurotransmitter (dopamine) depletion within the reward circuitry. Other theories posit that ‘opponent-process’ neural systems exist within the reward circuitry, mediating both positive and negative hedonic processes. In this view, craving results from functional dominance of neural systems mediating negative hedonic tone over those mediating positive hedonic tone. It is suggested that current neurophysiological, neurochemical, neuropharmacological and neuro-behavioral data appear to favor this latter view of the neurobiology of craving. It is further suggested that multiple craving states exist, with multiple neurotransmitter substrates. If true, such facts would imply that anti-drug-craving medication development programs predicated on a simple ‘hypo-dopaminergic’ concept of drug craving have little chance of success.