The choice between smoking, injecting or swallowing a drug influences the risk of addiction, as this determines both how much drug gets into the brain and how fast. Most animal studies on addiction focus on how much drug it takes to produce pathological drug use. How fast drugs get to the brain is generally ignored. A few studies have examined the influence of the speed of drug onset, but speed varied along with cumulative intake. Here we held average cumulative intake constant and determined whether variation in the speed of cocaine onset alone predicts outcome. Two groups of rats self-administered intravenous cocaine (0.25 mg/kg/injection) during daily sessions. Cocaine was available intermittently during each session. This produces the spikes and troughs in brain levels of cocaine thought to model how addicts take the drug. To vary the speed of cocaine onset, each injection was delivered over 5 s to one group, and over 90 s to the other. Average cumulative cocaine intake was the same in the two groups. However, rapid injections promoted robust psychomotor sensitization and potentiated incentive motivation for cocaine (0.063–0.25 mg/kg/injection). This addiction-relevant phenotype was accompanied by enhanced functional activity of metabotropic glutamate group II receptors (mGluR2/3s) in the prelimbic cortex and nucleus accumbens. Pharmacological activation of mGluR2/3s with LY379268 also preferentially decreased the motivation to take cocaine in rats previously exposed to rapid drug injections. Thus, varying the speed of drug onset can be used to parse the neurobiology of addiction from that of mere drug taking.