<p>Accelerometer (ACC) is one of the key payloads for all dedicated satellite gravity missions, and its accuracy seriously affects the quality of gravity field modeling. Accurately characterizing the relationship between accelerometer noise and the accuracy of gravity field modeling, is of great guiding significance for manufacturing accelerometers on one hand, and for mission parameter design on the other hand. In this study, we propose an analytical formula between the accelerometer radial observation noise and the geopotential coefficients errors from a novel perspective. Compared with the existing analytical methods and semi-analytical methods, our analytical formula reveals the role of frequency dependent accelerometer noise in the error propagation process from a more intuitive point of view. In order to test the accuracy of our analytical formula, we design a numerical simulation experiment to compare it with the least squares estimation (LSE) method, and analyze the influence of different spectral characteristics noise on gravity field modeling. The simulation results show that: (1) When the satellite ground tracks are sufficiently uniform and dense in the east–west direction, the results derived from our analytical formula are in good agreement with the classical LSE method, which demonstrates the reliability of our analytical formula. (<InternalRef RefID="Equ2">2</InternalRef>) When the noise with different spectral characteristics is input, error distribution patterns in geopotential coefficients also show different rules. These phenomena can be intuitively explained by our analytical theory. (3) Compared with the classical LSE method, the analytical method demonstrates superior in computational efficiency.</p>