Raw water with natural organic matters produces a variety of halogenic organic compounds after liquid chlorine or sodium hypochlorite disinfection, known as disinfection by-products (DBPs). Among them, the generation of trichloromethane (TCM) has attracted extensive attention. Factors affecting its formation including the organic content, pH, water temperature, chlorine dosage, disinfection contact time, and so on. According to survey, the scale of the water treatment plants in many cities and counties in northeast China is mainly small with a daily production capacity of less than 50 000 t/d. Moreover, due to the problems of testing capability and cost, it is impossible to detect TCM continuously in effluent water, and once the raw water quality changes, the safety of the effluent water quality will not be guaranteed. Therefore, the establishment of the prediction model between TCM and conventional parameters will help water treatment plants with no TCM detection abilities to predict the generation of TCM. In this study, taking a large water treatment plant in northeast China as an example, the contents of each parameter were detected using national standard methods. With the method of multiple linear regression analysis, the parameters which concern with the characteristics of raw water and disinfection processes, such as water temperature, pH, turbidity, potassium permanganate index, and chlorine dose, were added to the establishment of the TCM generation model. The multiple regression model provides a good prediction method for the generation of TCM in small and medium-sized water treatment plants with similar raw water characteristics and disinfection processes.