CPResearch started telephone interview from 1997. In 1998, CPResearch set up its Computer Assisted Telephone Interview (CATI). Now there are 30 stations in Beijing office. In order to satisfy clients request on CATI, CPResearch has set up its Web CATI from year 2004. 　 　 　 Web CATI 　 SurveyStar is a web-based CATI system, it is owned by CPResearch. In order to satisfy clients' increasing request on CATI, CPResearch launched its Web CATI in Shanghai and Guangzhou at year 2006, and will eventually expands into cities of the 2nd tier in the next several years. Up to now, CPResearch has 800 qualified CATI interviewers in Beijing, Shanghai, and Guangzhou. Eventually CPResearch will have several thousands qualified CATI interviewers all over China. SurveyStar consists of three separate server: server for dialing, for questionnaire management, and for database management. All three servers are connected into internet. Local computers log on the server for dialing, entering username and password, then can start interview. Web CATI will be fast because of large amount of interviewers, will be more accuracy because of interview in dialects, and will still high quality because of strict and transparent measures. Computer Assisted Telephone Interview (CATI) 　 　 　 CPResearch has totally 30 stations in Beijing CATI center. CATI has a number of advantages: 　 1) Standardized sampling and call-back procedures can be programmed in the system to insure uniformity of practice and more efficient calling routines. 2) Interview questins can be automatically modified to insert information already obtained and to phrase questions appropriately by such personal characteristics as gender and marital status. 　 3) Computer-controlled skip patterns permit far more complex interviews than are possible with paper and pencil forms. Questions can be designed to vary according to answers given earlier in the interview or even according to random numbers. Complex experiments can be integrated into the survey. 4) In-process data cleaning is a standard benefit, since many potential interviewer errors, such as missed questions or inappropriate skips, are virtually eliminated. Also, apparent discrepancies between responses may be automatically identified for probing during the course of the interview. 5) Tabulations and data files are available sooner because data entry and most manual editing and data cleaning steps are eliminated. For more complex studies, this may save weeks between the completion of interviewing and the beginning of analysis. 6) Automatic record-keeping, by date, time, sample segment, and interviewer, facilitates both interim and final reports on sampling outcomes and interviewer performance. Automatic timing of alternate question wordings and sequences can aid in the design of interviews that are less burdensome to respondents and more efficient for researchers.