A continuous spectrum water quality on-line monitoring signal processing method based on Hilbert-Huang transform (HHT) is proposed in this paper, which combines the micro-reagent water quality on-line monitoring technology of sequential injection. The modulation signal and spectrum curve of each intrinsic mode function (IMF) component of the original spectrum signal were obtained by empirical mode decomposition (EMD). The water sample data of different concentrations in the continuous spectrum on-line monitoring system was analyzed by the HHT model. The noise signal was excavated to realize the noise reduction processing, and the reconstruction of the continuous spectrum signal was realized after the noise reduction processing was completed. The research results show that this method can effectively reduce the noise of continuous spectrum signals according to different signal-to-noise characteristics of continuous spectrum, and has convenient use, fast processing speed, and high resolution in the time-frequency domain, which effectively improves the stability and accuracy of the micro-reagent continuous spectrum water quality on-line monitoring system.