To provide possible means to improve future satellite SST retrieval algorithms and to provide data users with error information, spatiotemporal bias patterns in the 1982--2002 NOAA operational AVHRR SST were characterized. At the start of one (NOAA-14) satellite mission, typical global average biases were about -0.05¿C, and gradually increased during the satellite mission. Normally global averages rarely exceeded 0.2¿C, due to cancellation of larger biases of opposite signs in different regions. Biases were inhomogeneous and locally exceeded 0.5¿C even in the 21-year time mean. Large irregular biases were due to stratospheric aerosols from volcanic eruptions. Seasonal biases were strongly related to local weather phenomena, such as seasonal dust aerosols and cloud covers. The findings suggest that future satellite SST retrieval algorithms should be time and space dependent.