The metabolic effects induced by antipsychotics in vitro depend on their action on the trafficking and biosynthesis of sterols and lipids. Previous research showed that antipsychotics with different adverse effects in patients cause similar alterations in vitro, suggesting the low clinical usefulness of cellular studies. Moreover, the inhibition of peripheral AMPK was suggested as potential aetiopathogenic mechanisms of olanzapine, and different effects on autophagy were reported for several antipsychotics. We thus assessed, in clinically-relevant culture conditions, the aetiopathogenic mechanisms of olanzapine, risperidone and ziprasidone, antipsychotics with respectively high, medium, low metabolic risk in patients, finding relevant differences among them. We highlighted that: olanzapine impairs lysosomal function affecting autophagy and autophagosome clearance, and increasing intracellular lipids and sterols; ziprasidone activates AMPK increasing the autophagic flux and reducing intracellular lipids; risperidone increases lipid accumulation, while it does not affect lysosomal function. These in vitro differences align with their different impact on patients. We also provided evidence that metformin add-on improved autophagy in olanzapine-treated cells and reduced lipid accumulation induced by both risperidone and olanzapine in an AMPK-dependent way; metformin also increased the production of bile acids to eliminate cholesterol accumulations caused by olanzapine. These results have different clinical implications. We demonstrated that antipsychotics with different metabolic impacts on patients actually have different mechanisms of action, thus supporting the possibility of a personalised antipsychotic treatment. Moreover, we found that metformin can fully revert the phenotype caused by risperidone but not the one caused by olanzapine, that still activates SREBP2.