BACKGROUND: Chaihu Injection (CI), which is widely used in treatment of febrile diseases, is an aqueous solution of Chaihu (Radix Bupleuri Chinensis) or Nanchaihu (Radix Bupleuri Scorzonerifolii) prepared by steam distillation. OBJECTIVE: This study aims at finding out the possible causes for adverse drug reaction or adverse event (ADR/AE) caused by CI and assessing its safety based on existing evidence. SEARCH STRATEGY: Manual search was not conducted. Electronic search was conducted by two authors in China National Knowledge Internet (CNKI) database and Chongqing VIP database (VIP). The search ended in June 30th, 2009. INCLUSION CRITERIA: Studies of ADR/AE induced by CI were collected comprehensively without considering language of literature and outcome indicators. Search results were not limited by patient's age, gender, race, primary disease, etc. Interventions were using CI alone or CI combined with other drugs (Chinese herbal medicine decoction or other drugs containing Chaihu were excluded). DATA EXTRACTION AND ANALYSIS: Two authors conducted data extraction independently. Microsoft Excel software was used to develop data extraction forms. Because of heterogeneity of the studies, only a descriptive analysis was conducted. RESULTS: Totally 83 studies with 203 cases were included in this review. Without the yield data and total amount of using, we cannot tell the incidence of ADR/AE induced by CI as well as assess the risk and safety of CI. The constituent ratio of severe cases was higher in children and old people than in other age groups. For most intramuscular cases, ADR/AE happened in 30 min after injection (constituent ratio of cumulative incidence in 30 min was 93.8%); for intravenously guttae patients, 4 cases of ADR/AE happened in the process of infusion; for first users, constituent ratio of cumulative incidence in 30 min and constituent ratio of cumulative incidence of severe cases in 30 min were higher than cases who had used CI before. Most ADRs/AEs were caused by incorrect use of CI, such as excessive doses (5 cases), intravenously guttae administration (6 cases), and violating incompatibility rules (7 cases). The incidence ratios of ADR and AE for severe and mild cases were 1.7:1 and 1.1:1, respectively; the ratios of the three relevant levels described as definitely related, most probably related and possibly related in the two types (severe and mild) of cases were 25:14:5 and 44:9:16, respectively. CONCLUSION: Present evidence with low level shows that incorrect use is the main cause of ADR/AE of CI. Whether CI is proper for children and old people still needs further research. Training for correct use of CI is necessary for medical workers. Much improvement in reporting ADR/AE based on "Recommendations for Reporting Adverse Drug Reactions and Adverse Events of Traditional Chinese Medicine" is in need.