A comparison has been undertaken between surface-observed total low- and high-cloud amount and retrievals from METEOSAT radiance data made using the cluster technique of Desbois et al. (1982). Ths aim of the study was to establish whether surface-observed cloud information could be ussfully exploited to benefit satellite-based cloud retrievals. Observations from 124 surface stations at 1200 UT for the 20-day period from July 22 to August 10, 1983, were compared with retrievals made from METEOSAT radiances measured at 1130 UT. The comparisons for total and low-cloud amount are made for France and southern Britain. The high-cloud amount comparison was limited to 34 stations in southern Britain. The location and time period were selected to coincide with one of the regions designated for special study in the International Satsllite Cloud Climatology Project (ISCCP) (Schiffer, 1982). For total cloud amount, 29% of the retrievals were fully in agreement with the surface observations and 64% of differences were within ¿1 okta (¿1 eighth of sky cover). In the case of layer cloud amounts, 64% of the low-cloud amount differences and 50% of the high-cloud amount differences were within ¿1 okta, although many of these successes (71% in the low-cloud amount) were for cases of totally clear or totally cloudy skies. Surface observations, which offer the only source of accurate low-cloud amount evaluation in any multilayered situation, were found to identify thin cirrus which was not detected by the satellite retrieval and to detect small gaps in cloud decks and small clouds missd by the satellite retrieval. In addition, cloud retrievals in coastal locations seemed to be more succsssfully accomplished by surface observers than by the satellite retrieval algorithm used here, which does not take into account land-sea partition. ¿American Geophysical Union 1987