In pregnancy, women are encouraged to cease smoking and limit caffeine intake. We employed objective definitions of smoking and caffeine exposure to assess their association with adverse outcomes. We conducted a case cohort study within the Pregnancy Outcome Prediction study to analyse maternal serum metabolomics in samples from 12, 20, 28 and 36 weeks of gestational age. Objective smoking status was defined based on detectable cotinine levels at each time point and objective caffeine exposure was based on tertiles of paraxanthine levels at each time point. We used logistic and linear regression to examine the association between cotinine, paraxanthine and the risk of pre-eclampsia, spontaneous pre-term birth (sPTB), fetal growth restriction (FGR), gestational diabetes mellitus and birthweight. There were 914 and 915 women in the smoking and caffeine analyses, respectively. Compared with no exposure to smoking, consistent exposure to smoking was associated with an increased risk of sPTB [adjusted odds ratio (aOR) = 2.58, 95% CI: 1.14 to 5.85)] and FGR (aOR = 4.07, 95% CI: 2.14 to 7.74) and lower birthweight (β = -387 g, 95% CI: -622 g to -153 g). On univariate analysis, consistently high levels of paraxanthine were associated with an increased risk of FGR but that association attenuated when adjusting for maternal characteristics and objective-but not self-reported-smoking status. Based on objective data, consistent exposure to smoking throughout pregnancy was strongly associated with sPTB and FGR. High levels of paraxanthine were not independently associated with any of the studied outcomes and were confounded by smoking.