ObjectiveTo examine the relationships between self-reported sleep characteristics and risk of incident vertebral fracture and hip fracture in women. DesignLongitudinal cohort study. SettingNurses’ Health Studies (NHS: 2002-2014, NHSII: 2001-2015). ParticipantsTotal 122,254 female registered nurses (46,129 NHS, 76,125 NHSII) without prior history of fracture. ExposureSleep was characterized by 4 sleep-related domains—sleep duration, sleep difficulty, snoring, and excessive daytime sleepiness—assessed by self-reported questionnaires. OutcomesSelf-reports of vertebral fracture were confirmed by medical record review and hip fracture was assessed by biennial questionnaires. ResultsOver 12-14 years of follow-up, 569 incident vertebral fracture cases (408 in NHS, 161 in NHSII) and 1,881 hip fracture cases (1,490 in NHS, 391 in NHSII) were documented. In the pooled analysis, the multivariable-adjusted HR (95% CI) for vertebral fracture was 1.20 (0.86, 1.66) for sleep duration ≤5 hours vs. 7 hours and 0.82 (0.60, 1.12) for ≥9 vs. 7 hours; 1.63 (0.93, 2.87) for sleep difficulties all-the-time vs. none/little-of-the-time (p-trend = 0.005); 1.47 (1.05, 2.05) for snoring every night/week vs. never/occasionally (p-trend = 0.03), and 2.20 (1.49, 3.25) for excessive daytime sleepiness daily vs. never (p-trend < 0.001). In contrast, associations were not observed with hip fracture risk. ConclusionPoorer sleep characteristics were associated with risk of vertebral fracture. Our study highlights the importance of multiple dimensions of sleep in the development of vertebral fractures. Further research is warranted to understand the role of sleep in bone health that may differ by fracture site, as well as sleep interventions that may reduce the risk of fracture.