ObjectiveTo evaluate the appropriateness, timeliness and perinatal outcomes of obstetric referrals compared to nonreferrals managed by caesarean section (CS) at a tertiary hospital.MethodsResearch assistants observed and recorded every delivery prospectively. Neonates were followed to discharge or up to 7 days, and outcomes were recorded. Proportions were used to describe baseline characteristics of the groups, and logistic regression was used to determine factors associated with CS in the referral group.ResultsA total of 10 353 deliveries occurred during the study period: 7.4% (768) were referrals and 92.6% (9585) were nonreferrals. The proportion of CS among referrals was 41.5% (319), nonreferrals 22.5% (2152) and overall 23.8% (2471). Compared to nonreferrals, referred women arrived at the second stage of labour with complications, ending up with emergency CS (p < 0.001). Factors associated with CS with odds ratio (OR) in the referral group were previous CS (11.06), abnormal foetal heart rate (4.96), breech presentation (4.61) and thick meconium (3.61). The referral group had higher proportions of stillbirths and neonatal admissions compared to nonreferrals (p < 0.001).ConclusionReferred women arrived late with complications ending with CSs and adverse perinatal outcomes. Earlier identification of complications and timely referral from lower‐level facilities may have the potential to prevent adverse outcomes.