Objective: To estimate the causal impact of small for gestational age (SGA) births on caesarean delivery, with and without trial of labour (TOL); and to quantify how much of the association is mediated through gestational age at delivery. Design: Cross-sectional analysis. Setting: Para 2 women who delivered non-anomalous, singleton live births from 22-44 weeks’ gestation in the US (2015-2018). Main outcomes and measures: Caesarean delivery with and without TOL. The exposure was SGA births (sex-specific birthweight <5th and <3rd percentiles for gestational age), and AGA births (10-89th percentile). We performed causal mediation analysis to determine the impact of gestational age at delivery (22-33, 34-36, 37-38, 39-40 and ≥41 weeks) as intermediate. Results: Of the 3,755,798 subjects, compared to AGA (29.6%), caesarean risks were higher for SGA <5th (34.3%) and SGA <3rd (36.4%) percentiles. For SGA <5th percentile, the adjusted excess risk of caesarean delivery without TOL had a “U” shaped association, with increased risk at preterm gestations, nadir at 39-40 weeks, and increased thereafter. The decomposition analysis revealed the driver of this excess risk was SGA births. The risk of caesarean delivery with TOL was highest <34 weeks’ gestation and was primarily an interaction effect. As gestation advanced, SGA births contributed proportionately greater to the risk. Associations were stronger for SGA <3rd percentile. Conclusions: Exposure to SGA drives high rates of prelabour caesareans and contributes to high risks of caesarean deliveries after TOL at >41 weeks gestation; a different mechanism drives high rates of caesareans after TOL at preterm gestations.