The quantitative [14C]-2-deoxyglucose autoradiographic method was utilized to assess regional cerebral metabolic rate for glucose (rCMRglc) in rat brain during withdrawal from cocaine self-administration. RCMRglc was determined in 62 regions from brains of naive rats which were placed into an empty operant chamber for 12 hr continuously, and rats trained to self-administer cocaine during 3 hr training sessions and subsequently placed into the operant chamber for 12 hr continuously with or without access to cocaine. Animals placed into the chamber without access to cocaine were examined 6 hr later, while animals allowed access to the 12 hr cocaine binge were examined either 6 or 72 hr post-cocaine. Metabolic activity was reduced during withdrawal in the nucleus accumbens, olfactory tubercle, islands of Calleja region, basolateral and central amygdaloid nuclei, medial septum, piriform and cingulate cortices, rostral caudatoputamen, entopeduncular nucleus and the adjacent lateral hypothalamus, somatosensory, auditory, and motor cortices compared to the naive state. These effects were usually more severe at 72 than at 6 hr after binge exposure, with intermediate values observed in cocaine trained animals without binge exposure. The response was negatively correlated with the amount of cocaine consumed during binge exposure in the striatum, olfactory tubercle, piriform, cingulate, somatosensory, and motor cortices. Thus, the amount of cocaine consumed can affect the extent of metabolic depression after sustained drug exposure. The pattern of regional effects suggests that mesolimbic and rostral extrapyramidal dopamine terminal regions and certain of their efferent pathways are preferentially affected during cocaine withdrawal. The reduction of basal metabolic rate observed in these brain regions during cocaine withdrawal may become more severe with time despite the apparent recovery of certain behavioral-motivational responses.