A new objective scheme has been developed for interpreting weather from direct output of numerical weather prediction models. The scheme produces seven categories of weather types : clear, fair, high cloud Overcast, cloudy, rain, snow and rain/snow mix, from predicted cloud amounts, one-hour accumulated precipitation amount, surface temperature and relative humidity. The scheme is applied to the JMA's operational mesoscale model and is verified against observed weather. The accuracy of the scheme is comparable with a 6-hour persistence forecast. This indicates a sufficient capability of the scheme for objective weather interpretation. Besides, weather distribution charts derived by the scheme have advantages in representing mesoscale disturbances compared to the traditional output of numerical models such as sea level pressure or 500 hPa height field. The precipitation-type forecast also demonstrated a good performance.