Mercury ion concentration is one of the key parameters affecting water quality environment, and a method for detecting mercury ion based on gold nanoparticle colorimetry was proposed. The gold nanoparticle colloidal suspension modified by bismuth reagent-Ⅱ was used to detect mercury ion. The mercury ion selectively coordinated with bismuth reagent-Ⅱ, which made the gold nanoparticle colloidal suspension anti agglomerate. There was a certain linear relationship between the concentration of mercury ion and the absorbance of the solution, the linear range was 0.05μg/L~1mg/L, and the relative standard deviation was 1.98%~3.52%. Under the experimental conditions, the characteristic wavelengths extracted from successive projection algorithm were pretreated by Savitzky-Golay first derivative. Through linear support vector machine regression model modeling and analysis, the model had good self prediction ability and practical prediction ability. The correlation coefficient of the calibration set was 0.9685, the root mean square error of calibration set was 0.0412mg/L, the correlation coefficient of the validation set was 0.9600, and the root mean square error of validation set was 0.0434mg/L. Therefore, the method based on gold nanoparticle colorimetry can provide effective support for improving the detection efficiency of mercury ion concentration. 