Caffeine is a commonly consumed drug during pregnancy with the potential to affect the developing fetus. Findings from previous studies have shown inconsistent results. We recruited a cohort of 2,643 pregnant women, aged 18-45 years, attending two UK maternity units between 8 and 12 weeks gestation from September 2003 to June 2006. We used a validated tool to assess caffeine intake at different stages of pregnancy and related this to late miscarriage and stillbirth, adjusting for confounders, including salivary cotinine as a biomarker of smoking status. There was a strong association between caffeine intake in the first trimester and subsequent late miscarriage and stillbirth, adjusting for confounders. Women whose pregnancies resulted in late miscarriage or stillbirth had higher caffeine intakes (geometric mean = 145 mg/day; 95% CI: 85-249) than those with live births (103 mg/day; 95% CI: 98-108). Compared to those consuming < 100 mg/day, odds ratios increased to 2.2 (95% CI: 0.7-7.1) for 100-199 mg/day, 1.7 (0.4-7.1) for 200-299 mg/day, and 5.1 (1.6-16.4) for 300+ mg/day (P (trend) = 0.004). Greater caffeine intake is associated with increases in late miscarriage and stillbirth. Despite remaining uncertainty in the strength of association, our study strengthens the observational evidence base on which current guidance is founded.