In this paper, a new algorithm for nitric oxide (NO) measurement based on multi-scale wavelet decomposition is proposed. The algorithm uses one-dimensional discrete multi-scale wavelet decomposition to analyze the transmission spectrum of flue gas. Through the different scales of decomposition, the relation between the absorption characteristics of NO gas and the reconstruction of wavelet coefficients is found. The DB9 wavelet can obtain the most obvious absorption characteristics of the NO gas. After calibration, the algorithm is used to measure the concentration of the mixture of NO and SO₂. The results show that the algorithm is efficient towards SO₂ absorption interference with a full-scale error less than ±0.4% in laboratory conditions. The algorithm was then applied to field measurements with an overall standard deviation of 8.37 mg m⁻³, which is better than 18.92 mg m⁻³ determined by the least squares method. Under the interference of desulfurization, the full-scale measurement errors are less than ±2%. In addition, the algorithm avoids using the reference spectrum, thereby reducing the aging of the light source effectively. It has strong anti-interference capability, good durability, and practicality in the continuous monitoring of NO.