Virtual reality (VR) technology can be a supporting tool to enhance mindfulness. Recently, many research using VR-based mindfulness (VBM) has been carried out in various psychiatric disorders but not in psychosis. We investigated safety and effects of virtual reality-based mindfulness (VBM) in patients with psychosis as a pilot study. Sixty-four patients were randomly assigned to VBM or to VR control. For VBM, education and meditation videos were provided. For VR control, 3-dimensional natural scenes were shown. Both programs consisted of 8 weekly sessions, each lasting about 30 min. Pre- and post-assessments were performed using the experiences questionnaire (EQ), psychotic symptom rating scales-delusion (PSYRATS-D), PSYRATS-auditory hallucinations (AH), motivation and pleasure scale-self rating (MAP-SR) and etc. The safety questionnaire was also surveyed after 1st and 8th session. Physiological measures such as skin conductance level (SCL), heart rate (HR) and RR interval, were collected during the VR interventions. Limited individuals participated in the safety questionnaire and physiological measures. All the results were presented in mean and standard deviation. We did not observe significant results in group x time interaction and main effects of group and time in the decentering and clinical scales. However, within group comparison showed that patients randomized to VBM showed increased decentering (p = 0.029) and decreased amount (p = 0.032) and duration of preoccupation (p = 0.016) in the PSYRATS-D. For the feelings and motivations about close caring relationships of the MAP-SR, we observed a significant group x time interaction (p = 0.027). The frequency of VR sickness was high but its severity was mild. There were significant differences only in HR over time in the VBM group (p = 0.01). These results suggest that VBM was not more effective in reducing decentering and psychiatric symptoms than VR control but its adversity was modest.