Proper identification of critical source areas (CSAs) is crucial for improving the efficiency of pollutant reduction and the economic viability of best management practices (BMPs) aimed at controlling non-point source pollution (NPSP). Different identification criteria, referring to different identification ratios, scales, and methods, affect the determination of CSAs, and then affect the spatial layout of BMPs. Nevertheless, few studies have optimized the BMPs placement in CSAs determined by different identification criteria and selected appropriate identification factors by comparing the effectiveness of BMPs schemes. To address this issue, a simulation-optimization method called simulation-based mixed-integer multi-objective non-linear programming (SMIMONLP) model was proposed by coupling Soil and Water Assessment Tool (SWAT) with mixed-integer multi-objective non-linear programming (MIMONLP). The method was presented to optimize BMPs placement in CSAs identified by different combinations of identification ratios, methods, and scales, while appropriate identification factors were selected by comparing the effectiveness of these BMPs optimization schemes. The effectiveness of these BMPs optimization schemes were evaluated by pollutant reduction, economic cost, and the satisfaction level of the decision maker with the objective function. The developed method was applied to a real case study in the Luan River Basin of North China. By comparing the economic cost and pollutant reduction of BMPs optimization schemes based on CSAs determined by different identification criteria, the results revealed that the economic cost difference of BMPs schemes could reach up to 5.9 times, 5.7 times, and 5.3 times due to the impact of CSAs identification ratio, scale and method, respectively. When other identification factors were the same, with the increase of pollutant reduction, the scenarios with a larger identification ratio, or using the Load Per Region Area Index (LPRAI) identification method, or in the hydrological response unit scale, tended to have lower economic cost and higher cost-effectiveness. The result of satisfaction level evualtion showed that higher environmental or economic satisfaction could be obtained from the identification criteria with the identification ratio of 20% and the identification scale of hydrological response unit, which could be priority selection criteria. This method can help decision makers select the BMPs schemes according to their preference and provide appropriate CSAs identification factors to optimize BMPs placement plans in watershed, improving the economic viability and pollutant reduction efficiency of NPSP