Advanced Very High Resolution Radiometer (AVHRR) aerosol optical thickness retrieval over the ocean is one of the two existing sources of long-term global satellite aerosol measurements (Total Ozone Mapping Spectrometer aerosol data set is the other). To make this 20-year historical data more useful for climate studies, the quality of the data (or the performance of the retrieval algorithm) has to be systematically evaluated. In this paper, as a continuation of our previous global validation effort, we present regional validation results for an AVHRR independent two-channel aerosol retrieval algorithm by comparing the retrievals with observations from the Aerosol Robotic Network (AERONET). The bias and the random errors of the retrieval algorithm applied to NOAA-14/AVHRR observations were determined and documented for key aerosol types (including biomass-burning, urban/industrial, desert dust, and marine). As a by-product of the validation, effective refractive indexes of the key aerosol types were also statistically determined through sensitivity analysis. The global and regional validations indicate that the new independent two-channel algorithm (with a globally unified aerosol model) performs well in the sense of the global mean. However, improvements are necessary to make the retrieval sensitive to aerosol types and to capture aerosol regional variations. The results will facilitate the utilization of long-term AVHRR aerosol products in climate studies and will provide guidance for improving aerosol retrievals from future NOAA satellite instruments.