Since 1970, NOAA (National Oceanic and Atmospheric Administration) AVHRR (Advanced Very High Resolution Radiometer) data have supported a broad range of environmental monitoring applications. In many of those applications, geometric corrections with high accuracy for NOAA AVHRR data are required for image transformation, registration and overlay. In order to use NOAA AVHRR data effectively, this study proposes a highly accurate geometric correction for NOAA data. Firstly, the elevation errors are calculated and corrected based on an elevation database. The elevation database is also used to divide NOAA data into flat and rough blocks. In addition to the GCPs (Ground Control Points) in the current GCP template database, more GCPs are generated based on the features of the coastline. Next, GCP template matching is used to calculate the residual errors of the blocks that match GCP templates. Based on these blocks, the residual errors of other flat and rough blocks are calculated by Affine and Radial Basic Function transformation respectively. According to the residual errors, all points in a NOAA image are moved to their correct positions. Finally, NOAA data is transformed from image into map coordinate system by bilinear interpolation. With the proposed method, the average error after correction is smaller than 0.15 pixels. The results proved that this is a highly accurate geometric correction method.