OBJECTIVE：To study the effects of protocol deviation on the results of clinical trials ，and to provide reference for rising the quality management of drug clinical trials. METHODS ：Blind data review forms for clinical trials of a contract research organzation （CRO） company in Guangzhou from 2010 to 2019 were collected to analyze general characteristics of protocol deviation，the situation of protocol deviation before and after the “722 announcement”（Announcement on Carrying Out Self-inspection and Verification of Drug Clinical Trial Data issued by CFDA on July 22，2015）as well as the effects of protocol deviation on full analysis set （FAS）population division. The suggestions were put forward. RESULTS ：A total of 45 trials were included，involving 454 centers，14 304 disease cases and 5 562 cases of protocol deviation. The most common types of protocol deviations were over-window ，violation of criteria of the inclusion and exclusion ，and drop-out ，which accounted for 36.88％， 20.71％ and 18.43％ respectively. There was no statistical significance in protocol deviation degree of clinical trials with different stages or drug types （P＞0.05）；there was significant difference in the degree of protocol deviations in clinical trials with different stages before and after the “722 announcement”（P＜0.05）；the incidence of deviations from over-window ，violation of cirteria of the inclusion and exclusion ，and medication compliance had increased after the “722 announcement”；82.07％ of cases with protocol deviations could enter FAS ，and the population who included in FAS but did not enter per protocol set （PPS）accounted for 53.99％ of the total deviation ，of which deviations from drop-out and combined medication accounted for 19.51％ and 4.29％ respectively. All cases with deviation from medication compliance did not enter PPS. CONCLUSIONS ：Drop-out，violation of criteria of the inclusion and exclusion ，and over-window are the main factors that cause clinical trial protocol deviations. The “722 announcement”played a certain role on improving the quality management awareness of the personnel in drug clinical trial. Appropriate statistical methods should be selected to control bias ，and to strengthen the quality management of drug clinical trials and reduce protocol deviations ，by paying attention to trial design， staff training ， institutional management and 85223869。