Objective To define a trimester-specific reference interval for C-reactive protein (CRP) in healthy pregnant women, and to evaluate its accuracy for diagnosing infection. Design Retrospective cross-sectional and diagnostic accuracy study. Setting Tertiary hospital in Oxford, UK. Population Development cohort: 315 pregnant women from the Oxford Pregnancy Biobank with uncomplicated pregnancies, with longitudinal sampling in each trimester. Evaluation cohort: 50 pregnant women with suspected chorioamnionitis. Methods and main outcome measures We calculated a 95% reference interval for CRP in each trimester and evaluated its diagnostic accuracy for infection compared to that from current guidance (≤7 mg/L). Results Of the 315 healthy pregnant women in our study, concentrations of CRP were substantially higher than those in most non-pregnant populations. The reference intervals in each trimester were similar, with an upper reference limit of 18 mg/L. CRP increased log-linearly with body mass index in all trimesters (p<0.001). The sensitivity and specificity of CRP for diagnosing chorioamnionitis were 80% and 86%, respectively. The overall diagnostic accuracy using the pregnancy-specific reference interval was significantly greater than that of the existing standard (p=0.002). Conclusions A pregnancy-specific reference interval for CRP should be used to optimise diagnostic accuracy for infection in pregnant women. Chorioamnionitis was used as example of a localised infection with well-defined outcomes, but pregnancy-specific upper reference limits for CRP should be considered in any clinical setting including pregnant women.