Antipsychotic medications are notorious for their associations with toxicity concerns at different systemic levels and their many side effects, varying from common and unpleasant issues to disfiguring medication-induced conditions (tardive dyskinesia, obesity) and even life-threatening disorders (myocarditis, agranulocytosis). Our study focuses on in silico design of novel drug prototypes against schizophrenia, a disorder with a complex pathological mechanism involving the dysregulation of different pathways. For this particular design, the receptors of dopamine and serotonin were used as molecular targets. A combinatorial library was computationally generated based on scaffolds from four categories of compounds: one worldwide approved typical antipsychotic (Chlorpromazine – CHL), two worldwide approved atypical antipsychotics (Risperidone – RIS, Haloperidol – HAL), one atypical antipsychotic approved only in Japan (Emonapride – EMO), and one compound used in pharmacological research (Eticlopride – ETI). The combinatorial library (scaffolds' donators included) was screened for lead-likeness, drug-likeness, activity at the central nervous system (CNS) and ADMET properties. Two good drug candidates (virtual derivatives of ETI) were identified, with activity at the CNS level, free of toxic substructure and without additional toxicity concerns. One virtual derivative of HAL was found with activity at the CNS level, with an inherited low-risk toxic substructure, but without other additional toxicity concerns acquired from its scaffold donor. The selected virtual derivatives were screened against the primary targets of the scaffolds' donors using molecular docking. What was found is that the virtual derivatives had better binding affinities than their scaffolds' donors against some of the dopamine receptors.