A turbidity disturbance compensation algorithm based on the Mie scattering theory for water quality measurement by ultraviolet-visible spectroscopy is studied in order to reduce the turbidity disturbance caused by light scattering from suspended particles in the waterbody. The disturbance leads to a nonlinear lifting of the spectrum and results in significant decrease of the measurement accuracy. By the total light scattering method based on the Mie scattering theory, particle size distribution of the water sample can be reconstructed through the ultraviolet-visible spectrum in the region from 450 nm to 1100 nm, which is mainly related to the particle scattering. Then, with the secondary inversionary of the particle size distribution, extinction spectrum caused by particle scattering in the wavelength region of 220-450 nm which is the characteristic spectrum region of the organic pollutant can be estimated. So the accurate compensation of the turbidity distribution is realized. Experimental results show that a good effect of turbidity compensation for different water samples is achieved by the algorithm without a lot of prior experimental data. Calculation accuracy of the chemical oxygen demand has been greatly improved. It is of practical value for increasing the accuracy, flexibility and adaptability in the ultraviolet-visible spectroscopy for water soluble organic measurement.