AbstractBackgroundGestational weight gain may be a modifiable contributor to infant health outcomes, but the effect of gestational duration on gestational weight gain has limited the identification of optimal weight gain ranges. Recently developed z‐score and percentile charts can be used to classify gestational weight gain independent of gestational duration. However, racial/ethnic variation in gestational weight gain and the possibility that optimal weight gain differs among racial/ethnic groups could affect generalizability of the z‐score charts. The objectives of this study were (1) to apply the weight gain z‐score charts in two different U.S. populations as an assessment of generalisability and (2) to determine whether race/ethnicity modifies the weight gain range associated with minimal risk of preterm birth.MethodsThe study sample included over 4 million live, singleton births in California (2007–2012) and Pennsylvania (2003–2013). We implemented a noninferiority margin approach in stratified subgroups to determine weight gain ranges for which the adjusted predicted marginal risk of preterm birth (gestation <37 weeks) was within 1 or 2 percentage points of the lowest observed risk.ResultsThere were minimal differences in the optimal ranges of gestational weight gain between California and Pennsylvania births, and among several racial/ethnic groups in California. The optimal ranges decreased as severity of prepregnancy obesity increased in all groups.ConclusionsThe findings support the use of weight gain z‐score charts for studying gestational age‐dependent outcomes in diverse U.S. populations and do not support weight gain recommendations tailored to race/ethnicity.