The mechanism underlying the serotoninergic neurotoxicity of some amphetamine derivatives, such as p-chloroamphetamine (pCA) and 3,4-methylenedioxymethamphetamine (MDMA), is still debated. Their main acute effect, serotonin (5-HT) release from nerve endings, involves their interaction with 5-HT transporters (SERTs), as substrates. Although this interaction is required for the neurotoxic effects, 5-HT release alone may not be sufficient to induce long-term 5-HT deficits. Some non-neurotoxic compounds, including p-methylthioamphetamine (MTA) and 1-(m-chlorophenyl)piperazine (mCPP), have 5-HT releasing properties in vivo and in brain slices comparable to that of neurotoxic amphetamine derivatives. We measured 5-HT release in superfused rat brain synaptosomes preloaded with [3H]5-HT, a model that distinguishes a releasing effect from reuptake inhibition. MTA and mCPP induced much lower release than pCA and MDMA. The striking difference between our findings in synaptosomes and those obtained in vivo or in brain slices is probably related to a different compartmentalisation of 5-HT in the different experimental models. Studies in synaptosomes, where the vesicular storage of 5-HT is predominant, could therefore bring to light differences between neurotoxic and non-neurotoxic 5-HT releasing agents which cannot be appreciated in other experimental models and might be useful to identify the mechanisms responsible for the neurotoxicity induced by amphetamine derivatives.