About 30% of people aged over 65 years living in the community fall at least once a year; the fall rate is even higher in nursing homes. Many of these falls lead to fractures, the most serious type being hip fractures. Whether or not the use of various medications causes falls and fractures has been the subject of more than 50 observational epidemiological studies, most published after 1988. Few of these studies were specifically designed to investigate the associations between medication use and falls, so most have methodological limitations. Uncontrolled confounding factors are a particular problem. It is important to recognise these flaws in study design when reviewing the scientific literature on the link between medications and falls and fractures. Patients taking psychotropic medications appear to have about a 2-fold increased risk of falls and fractures, compared with individuals not taking these drugs. The strongest evidence is for antidepressants; nearly every relevant study has found an association with falls. The relative effect of selective serotonin (5-hydroxytryptamine; 5-HT) reuptake inhibitors and tricyclic antidepressants on falls is still unknown. The effect on the risk of falling of short-acting versus long-acting benzodiazepines is not straightforward: it may be the benzodiazepine dosage that matters, rather than the drug half-life. Whether or not any medications used to treat cardiovascular problems cause falls remains unclear. Current evidence suggests that diuretics, in general, do not cause falls and that thiazide diuretics, in particular, may help prevent fractures by slowing the development of osteoporosis. Some studies have found that use of nonsteroidal anti-inflammatory drugs is associated with falling; this deserves further study. Reducing the use of psychotropic drugs by residents of nursing homes should be a high priority for physicians, pharmacists and nursing staff. Psychotropic drug use should also be kept to a minimum among older people living in the community.