Objective To assess the influence of body mass index (BMI) on the performance of epidural analgesia in labour and the subsequent mode of delivery. Design A retrospective audit of prospectively collected quality assurance data. Setting The delivery suite of Leeds General Infirmary, Leeds, UK. This is a 4500-delivery teaching hospital unit. Population All women receiving epidural analgesia during labour in our unit between April 1997 and December 2005. Methods Epidural recipients were divided into BMI groups according to World Health Organization (WHO) categories and compared for indices of epidural performance and mode of delivery. Main outcome measures Midwife and patient satisfaction scores with epidural analgesia, epidural resite rates, and mode of delivery. Results Data from 13 299 epidural recipients were analysed. Using WHO definitions, 22.8% were of normal body mass, 41.9% were overweight, 31.9% obese, and 3.4% morbidly obese. Epidurals were more likely to fail as BMI increased, as judged by midwife satisfaction scores (P < 0.001) and epidural resite rates (P < 0.01). This trend was not seen for maternal satisfaction scores using the WHO BMI categories. However, if women with BMI below 30 kg/m2 were grouped together, a significant trend was found (P < 0.01). BMI had no influence on vaginal instrumental deliveries, but caesarean section rates rose from 11.5% in women of normal BMI to 29.2% in the morbidly obese women (P < 0.001). Conclusions Obesity increases the incidence of analgesic failure and the need for resite of epidurals. The caesarean section rate among epidural recipients increases dramatically as BMI rises.