Passive FTIR infrared remote sensing technology is useful in gas leak detection. The detection limit of trace gas is related to the instrument’s signal-to-noise ratio (SNR). The instrumental SNR is related to measurement parameters, such as spectral resolution, sampling frequency, integration time, and average times. How to optimize the combination of parameters to achieve the best signal-to-noise ratio in combination with actual applications currently lacks system analysis. This paper the oretically analyzes the relationship between these parameters and SNR and categorizes them into three aspects: (1) In terms of spectral resolution, the conclusion, cited from Roland Harig, is that SNR stays the same when the spectral resolution is lower than the full width at half maximum (FWHM), but in order to avoid interference of background gases, the resolution mustavoid too low and is appropriately set in practical application; (2) In terms of the sampling frequency, sampling frequency reduction can reduce calculation cost and narrow the spectral range, but the sampling frequency and spectral range are not related with SNR; (3) In terms of integration time and multiple spectra averaging, under the same time conditions, multiple interferogram acquisitions will introduce a sampling error in zero path difference, making the noise larger than the theoretical calculation, so SNR obtained by long integration time is better than multiple averaging. Conducted sulfur hexafluoride (SF6) leak detection experiment, the 4 cm-1 resolution is selected according to SF6 spectral feature, and 20 kHz sampling frequency is selected by taking into account SNR and detection time, and FTIR inspection system is used to locate the leakage point in the transform substation, which proves the effectiveness of this method.