This article tries to summarize and synthetically remark several basic aspects and common inherent characteristics of atmospheric data assimilation. Such a logic line is highlighted from “what’s data assimilation problem”, to “how is data assimilated”, and to “what are the inherent requirements for the resulting analysis”. The logic correspondingly involves “inverse problem and a priori information”, “input, weight, propagation and smoothing of observation information”, and “requirements of accuracy and consistency for the analysis”. Also, we lay emphasis on some common inherent characteristics, including “equivalences between variational method and optimal interpolation method and between fourdimensional variational method and extended Kalman filter method”, “the dominant role of mathematics in assimilation methods”, “multidisciplinary knowledge connotated in assimilation methods” and “system engineering required in operational assimilation systems”.