The invention discloses a COD (Chemical Oxygen Demand) ultraviolet spectrum online detection optimization method based on a neural network and belongs to the technical field of detection of organic matters in a water body. The method comprises the following steps: measuring the turbidity, suspension solids, conductivity and pH (Potential of Hydrogen) value of a water sample to be detected; measuring COD value of the water sample to be detected by utilizing an ultraviolet spectrum method and taking the COD value as COD initial value of the water sample to be detected; taking the turbidity, suspension solids, conductivity, pH value and COD initial value of the water sample to be detected as input vectors of a trained BP neural network, so as to obtain COD optimized value of the water sample to be detected. According to the COD ultraviolet spectrum online detection optimization method, high-precision neural network modeling can be carried out according to various influence factors of the water body so as to improve the precision of online measurement of the ultraviolet spectrum method. The influence factors considered by the COD ultraviolet spectrum online detection optimization method comprise the turbidity, suspension solids (SS), conductivity and pH value of the water sample. The factors influence the accuracy of measuring the COD value by the ultraviolet spectrum method. By modeling the neural network, the influences, caused by the factors, on COD measured value can be accurately estimated so that the more accurate COD optimized value is obtained.