Lone-term satellite observations, such as Advanced Very High Resolution Radiometer (AVHRR), provide an irreplaceable means in monitoring Earth system through a series of satellites. However, to be able to detect the signal related to climate change, one of the critical requirements is the consistency and stability of calibration among the satellites. Applying Simultaneous Nadir Overpass (SNOs) method (Cao et al., 2002)., we fully accessed instrument-related consistency of AVHRR measurements covering all channels (from visible to IR) and time period from 1978 to 2003. It is seen that the inter-satellite biases in visible channels (channel 1 and 2) show larger inconsistency among satellites especially between NOAA-14 and NOAA-12. The inconsistency is shown as both the large bias and trend in the biases, mostly due to the lack of onboard calibration. Comparatively, the biases in IR channels, i.e., channel 4 and 5 are generally smaller, there are within ± 1 k. However, the difference in the magnitude of the biases among satellites and the dependence of biases on the scene temperature may affect the quality of long term trend derived from such dataset. Analyses of bias root causes indicate that the effect from the difference in Spectral Response Function may not be large enough to account for the observed biases.