A recent study reported a decreased risk of fractures among lithium users. We conducted a case-control study within the UK General Practice Research Database, comparing never, ever, current, recent and past lithium use in 231,778 fracture cases to matched controls. In addition, the risk of fractures was assessed in relation to cumulative duration of use and time since discontinuation. Current use of lithium was associated with a decreased risk of fractures (adjusted odds ratio [OR]=0.75, 95% confidence interval [CI]=0.64-0.88), which did not vary with cumulative duration of use. Among past users an increased risk of fractures was observed (adjusted OR=1.35, 95% CI=1.01-1.79), increasing with time since discontinuation. Our results support the role of the underlying mental disorders in the aetiology of fractures and do not support a pharmacological effect of lithium based on lack of an association with cumulative duration of use.