In order to evaluate the effect of refined multi-model objective consensus forecasting service products (OCF) temperature forecasts which are applied in public weather service, and analyze the causes of forecast errors, this paper makes objective verification on OCF daily maximum and minimum temperature forecasts in China, focuses on large error days with high service impact and the typical case： temperature-drop days, and also makes comparison among OCF, ECMWF and NCEP. The results show that OCF daily maximum and minimum temperature forecasts perform better than models in the consensus generally, and the forecast accuracy is higher in summer but lower in winter. OCF enlarges the range of daily temperature variation and effectively reduces forecast errors. OCF has fewer large error days than models in the consensus, but shows larger errors in 2-3 d forecasting periods and winter half year. The large error days of OCF are related to models in the consensus and obvious temperature-drop. It is found that the forecast performances of OCF, ECMWF and NCEP decline in temperature-drop days, and the error of OCF daily maximum temperature forecast increases rapidly. In temperature-drop days, OCF effectively corrects daily minimum temperature and daily maximum temperature in non-temperature-drop areas, but the daily maximum temperature forecast performs with obvious positive error in temperature-drop areas. Finally, based on analysis above, the improvement direction for OCF consensus methods is proposed. The process verification is conducive to discovering the defects of objective forecasts and methods of temperature consensus and correction.