Benzodiazepine receptor agonists (BZDRAs) have been associated with an increased risk of falls in the elderly. However, the association between the elimination half-life (t 1/2) of BZDRAs and the diŠerence between benzodiazepines (BZDs) and non-benzodiazepines (Z-drugs) has not been clariˆed. By conducting a meta-analysis of observational studies, we compared the risk of falls with respect to 1) short-acting BZDRAs (t 1/2 ＜ 12 h) vs. long-acting BZDRAs (t 1/2 12 h) and 2) BZDs vs. Z-drugs in elderly patients. Data were retrieved from MEDLINE, the Cochrane Library, and Igaku Chuo Zasshi. In total, 13 observational studies from 12 articles were included in our study (short-acting BZDRAs, n ＝ 12; long-acting BZDRAs, n ＝ 9; BZDs, n ＝ 13; Z-drugs, n ＝ 7). The risk of falls was signiˆcantly increased by the use of short-acting BZDRAs [Odds ratio (OR) (95％ Conˆdence interval (CI)): 2.00 (1.46 2.73)], long-acting BZDRAs [OR (95％CI): 2.16 (1.61 2.89)], BZDs [OR (95％CI): 1.67 (1.31 2.13)], and Z-drugs [OR (95％CI): 2.42 (1.35 4.34)] compared to the risk in BZDRAs non-users. The increased risk of falls in elderly patients was similar in each group and unrelated to t 1/2. This study suggested that all BZDRAs including Z-drugs should be avoided in elderly patients.