A retrospective analysis was performed on 19,259 deliveries that occurred in our institution from January 2000 to December 2002. Anesthesia records and quality assurance data sheets were reviewed for the characteristics and failure rates of neuraxial blocks performed for labor analgesia and anesthesia. The neuraxial labor analgesia rate was 75% and the overall failure rate was 12%. After adequate analgesia from initial placement, 6.8% of patients had subsequent inadequate analgesia during labor that required epidural catheter replacement. Ultimately 98.8% of all patients received adequate analgesia even though 1.5% of patients had multiple replacements. Six percent of epidural catheters had initial intravenous placement but 46% were made functional by simple manipulations without higher subsequent failure. Unintended dural puncture occurred in 1.2% of labor neuraxial analgesia. The incidences of overall failure, intravenous epidural catheter, wet tap, inadequate epidural analgesia and catheter replacement were lower in patients receiving combined spinal-epidural versus epidural analgesia. For cesarean section, 7.1% of pre-existing labor epidural catheters failed and 4.3% of patients required conversion to general anesthesia. Spinal anesthesia for cesarean section had a lower failure rate of 2.7%, with 1.2% of the patients requiring general anesthesia. The overall use of general anesthesia decreased from 8% to 4.3% over the three-year period. Furthermore, regional anesthesia was used in 93.5% of cesarean deliveries with no anesthetic-related mortalities. Future investigations should identify acceptable international standards, risk factors associated with failure and methods to reduce failure before cesarean section.