Copy number variations in theARHGAP10gene encoding Rho GTPase–activating protein 10 are significantly associated with schizophrenia. ARHGAP10 negatively regulates RhoA/Rho-kinase (ROCK) signaling. We previously demonstrated that fasudil, a non-selective ROCK inhibitor, exhibited antipsychotic-like effects in several mouse models of schizophrenia. ROCK has two subtypes, ROCK1 and ROCK2. ROCK1 is mainly expressed in the thymus and blood, while ROCK2 is predominantly expressed in the brain. Therefore, it is expected that like fasudil, selective ROCK2 inhibitors will exhibit antipsychotic-like effects, accompanied by a lower incidence of adverse effects due to ROCK1 inhibition. Here, we used genetic and pharmacological models of schizophrenia to investigate whether the selective ROCK2 inhibitor KD025 would show antipsychotic-like effects with a favorable adverse effect profile. Oral administration of KD025 suppressed the abnormal increase in the phosphorylation level of myosin phosphatase–targeting subunit 1, a substrate of ROCK, and ameliorated the decreased spine density of layer 2/3 pyramidal neurons in the medial prefrontal cortex ofArhgap10S490P/NHEJ mice. Furthermore, KD025 mitigated the methamphetamine-induced impairment of visual discrimination (VD) inArhgap10S490P/NHEJ and wild-type mice. KD025 also reduced MK-801–induced impairments of VD, novel object recognition, and hyperlocomotion. Regarding side effects that are commonly seen with typical antipsychotics, KD025 did not affect systolic blood pressure and did not induce extrapyramidal symptoms, hyperprolactinemia, or hyperglycemia at the effective dosage in naive wild-type mice. Taken together, KD025 shows antipsychotic-like effects with a favorable adverse effect profile in genetic and pharmacological mouse models of schizophrenia.