A simple objective cloud type classification method has been developed, based on split-window measurements of the Advanced Very High Resolution Radiometer on board the NOAA 7 satellite. Brightness temperature difference between the split-window data is a good parameter for the detection of cirrus and blackbody clouds. Two-dimensional histograms of brightness temperature of the 11μm channel and the brightness temperature difference between the split-window data over (64 km)2 subareas are constructed. By selecting appropriate thresholds in the two-dimensional histogram, cirrus, dense cirrus, cumulonimbus, and cumulus clouds are classified over the tropical ocean. Cloud type classification maps were generated by this method for the western Pacific Ocean and were compared with the nephanalysis chart constructed at the Japan Meteorological Satellite Center from GMS data collected within 1 hour of the NOAA 7 observations. The comparison shows reasonable agreement. Fractional cloud cover for cirrus over each (64 km)2 subarea is calculated as the ratio of the number of samples which belong to the cirrus cloud type in the two-dimensional histogram to the number of total samples in the subarea. Fractional cloud cover estimations for cumulonimbus and low-level cumulus are also presented. ¿American Geophysical Union 1987