Cocaine has actions in the peripheral nervous system that reliably precede--and thus predict--its soon-to-follow central rewarding effects. In cocaine-experienced animals, the peripheral cocaine signal is relayed to the central nervous system, triggering excitatory input to the ventral tegmental origin of the mesocorticolimbic dopamine system, the system that mediates the rewarding effects of the drug. We used cocaine methiodide, a cocaine analog that does not cross the blood-brain barrier, to isolate the peripheral actions of cocaine and determine their central and behavioral effects in animals first trained to lever-press for cocaine hydrochloride (the centrally acting and abused form of the drug). We first confirmed with fast-scan cyclic voltammetry that cocaine methiodide causes rapid dopamine release from dopamine terminals in cocaine hydrochloride-trained rats. We then compared the ability of cocaine hydrochloride and cocaine methiodide to establish conditioned place preferences in rats with self-administration experience. While cocaine hydrochloride established stronger place preferences, cocaine methiodide was also effective and its effectiveness increased (incubated) over weeks of cocaine abstinence. Cocaine self-administration was extinguished when cocaine methiodide or saline was substituted for cocaine hydrochloride in the intravenous self-administration paradigm, but cocaine hydrochloride and cocaine methiodide each reinstated non-rewarded lever-pressing after extinction. Rats extinguished by cocaine methiodide substitution showed weaker cocaine-induced reinstatement than rats extinguished by saline substitution. These findings suggest that the conditioned peripheral effects of cocaine can contribute significantly to cocaine-induced (but not stress-induced) cocaine craving, and also suggest the cocaine cue as an important target for cue-exposure therapies for cocaine addiction.