Intra-individual variability (IIV) has received recent attention as an indicator of the stability of cognitive functioning that may outperform mean performance in reflecting putative neurobiological abnormalities. Increased IIV is regarded as a core deficit in schizophrenia patients; however, whether this deficit is present in the prodromal phase before the onset of schizophrenia has not been well established. In the present study, we investigated IIV using the stop-signal paradigm in at-risk mental state (ARMS) individuals and in schizophrenia patients. The study included 27 ARMS subjects, 37 schizophrenia patients, and 38 normal controls. The stop-signal task was administered to assess IIV and response inhibition. IIV was estimated by calculating the standard deviation across sub-blocks for the three groups. We observed increased IIV in ARMS subjects and schizophrenia patients compared with normal controls in both the "stop" and the "go" processes even though the mean response inhibition performances were not impaired in the ARMS group. Schizophrenia patients showed impaired response inhibition that was associated with the severity of negative symptoms. Our findings suggest that the analysis of IIV may identify cognitive and clinical features of ARMS that are not detectable by conventional mean performance analysis. The unstable response patterns associated with ARMS may originate from abnormal processing in neural systems caused by alterations in the integrity of functional brain networks and dopamine neuromodulation.