Abstract: The Second State Pollution Source Survey has published the survey bulletin in June 2020, which means that over three-year national basic surveys has entered the final stage successfully. More than 3.58 million units and self-employed industrial or agricultural households participated in data filling during the household survey. As far as data collection and result application were concerned, a timely review of the practices and methods used in the past processes could have great significance to the use and evaluation of the results. However, enough attentions were also required to find and summarize the common problems that occurred in data filling in the meantime. In order to draw a comprehensive conclusion of the household survey results, and to reduce and avoid the above-mentioned problems in the Second State Pollution Source Survey, this paper reviewed and analyzed the whole process of the survey combination with preparation before the survey, data verification and calculation procedure after the survey. Single data source, backward data collection methods, lack of audit technology application were obvious problems found in the process. In the course of the survey, we analyzed the advantages and disadvantages of the three modes of operation to make the following routes: according to the geographic distribution of the survey target, according to the type and scale of the survey target, based on both the geographic distribution and the type and scale. The first route is suitable for the areas where the objects are evenly distributed and balanced resources. The second route is restricted by time and other factors. The last route need well organized and planned before household survey. After that, some valid measures were expected to put forward according to the key issues that could affect the accuracy and availability of collected data, including data source selection, data collection reliability, audit technology application. Through the selection of past examples among primary survey agencies along with real cases analysis and literature collection, useful advice and measures have arisen as follows. Firstly, simplifying indicators that need to fill in the basic information forms with simultaneous examination and verification automatically. Besides, placing the logical framework of judgment by using technological examination in front of submitting which could avoid a majority of errors. If possible, combining other related resources could take advantage in optimizing the process under certain circumstances. Based on the above-mentioned main measures, the results obtained from that would bring more benefit for carrying out related information collection, improving the data quality, and raising the standard of environmental management in the future.