Chronic exposure to cannabinoids was shown to induce long lasting impairment of learning and memory, which was accompanied by morphological damage to the brain. On the other hand, several studies have shown that cannabinoids can protect from various brain traumas. This enigmatic dualism is explained herein by a comprehensive hypothesis, which is based on our recent in vitro studies and on pharmacokinetic in vivo considerations. The hypothesis predicts that low concentrations of cannabinoids will be neurotoxic while high concentrations of the drugs will protect from neuronal damage, and suggests that chronic administration of cannabinoids will induce neuronal death, while their acute administration will protect the brain. We further propose straight forward experiments, both in vivo (animal models for brain damage) and in vitro (cell death in neuronal cultures) to verify this hypothesis. The outcome of these experiments may have practical applications when considering the use of cannabinoids as therapeutic agents and in evaluating the consequences of their use as recreational drugs.