Validation studies indicate that in cloud free conditions the along-track scanning radiometer (ATSR) on the ERS 1 satellite is measuring global sea surface temperature (SST) to a point relative precision of better than 0.3 ¿C. However, recent reports show that the presence of nighttime remnant cloud contamination is affecting ATSR SST precision in certain geographical areas. We modify the filtering scheme described by Jones et al. <1996> for removing remnant cloudiness in ATSR 0.5¿ spatially averaged sea surface temperature (ASST) data from the South Atlantic and apply this scheme to global ASST data (version 500) for the whole of 1992, 1993, and 1994. We find that globally 4.65% of the ASST data is cloud contaminated and that remnant cloud removal substantially reduces differences between day and night SST signals. The cloud contamination is seasonal and is linked to the occurrence of marine stratiform (fog, stratus, and stratocumulus) cloud types. Such cloud is low-lying and uniform and thus difficult to detect with conventional cloud-clearing techniques. Our effective global removal of remnant cloudiness will improve the accuracy of the ASST data and thus benefit the use of the ATSR in the early detection of climate change. ¿ American Geophysical Union 1996