A robust performance evaluation method for vapor cells used in magnetometers is proposed in this work. The performance of the vapor cell determines the magnetic measurement sensitivity, which is the core parameter of the magnetometer. After establishing the relationship between the intrinsic sensitivity and the total relaxation rate, the total relaxation rate of the vapor cell can be obtained to represent the intrinsic sensitivity of the magnetometer by fitting the parameters of the magnetic resonance experiments. The total relaxation rate measurement method based on the magnetic resonance experiment proposed in this work is robust and insensitive to ambient noise. Experiments show that compared with the conventional sensitivity measurement, the total relaxation rate affected by magnetic noise below 0.9 nT, pump light frequency noise below 1.5 GHz, pump light power noise below 9%, probe light power noise below 3%, and 150±3 ℃ temperature fluctuation deviates less than 2% from the noise-free situation. This robust performance evaluation method for vapor cells is conducive to construct a multi-channel high spatial resolution cardio-encephalography system.