About an hour from asking, and about 1 in 8 to 1 in 4 do not work well
In the only careful timing study, the median from request to comfort was 65 minutes, and published failure rates run from 12% to 23% depending on what counts as failure.
Two practical things about the epidural that are rarely said in advance: it takes time, and it does not always work.
The time. A prospective study recorded 324 requests for an epidural at two large Australian maternity units. Every step was timed.
The authors put it plainly: about an hour from asking to comfort. That is Australian data from 2009 and 2010, in two hospitals with anesthesiologists on site. No equivalent American figure has been published, so your hospital's wait may be shorter or longer, and it will differ by time of day. Ask them for their own.
You may also be shown targets rather than measurements. A 2025 expert consensus proposed quality goals for obstetric anesthesia, including at least 90% of eligible patients receiving neuraxial analgesia within 30 minutes of request. Those are goals a group of specialists wrote down as something to aim at. They are not a record of what happens, and nobody should quote them to you as if they were.
The failures. There is no agreed definition of a failed epidural, and that single fact explains the whole range of published numbers.
| Study | What was counted | Rate |
|---|---|---|
| 19,259 US deliveries | The hospital's own all-cause definition | 12% |
| 4,240 US anesthesia records | Catheter had to be replaced | 13.1% |
| 1,521 UK epidurals | A definition agreed by a specialist panel, including pain relief 45 minutes after the start | 23% |
The fair sentence is: depending on how failure is defined, published rates of an epidural that does not work well enough run from about 1 in 8 to about 1 in 4. The high number is not a worse hospital; it is a stricter ruler.
Some more specific figures from the largest of those studies, all of them American:
6.8% of women who got good relief at first later needed the catheter replaced. 1.5% needed more than one replacement. In the end, 98.8% of women received adequate pain relief. Accidental puncture of the covering of the spinal cord happened in 1.2%. In the study that counted replacements, satisfaction was still above 98%.
If a cesarean becomes necessary, the epidural already in your back is usually topped up rather than replaced with a general anesthetic. In that same 19,259-delivery series, the existing labor catheter failed for the cesarean in 7.1% of cases, and 4.3% of women needed a general anesthetic instead. Regional anesthesia was used for 93.5% of cesareans, with no anesthetic-related deaths in the series.
A pooled analysis of 13 studies and 8,628 women found three things that make that conversion more likely to fail: more clinician-given top-ups during labor (odds ratio 3.2, 95% confidence interval 1.8 to 5.5), greater urgency of the cesarean (odds ratio 40.4, 95% confidence interval 8.8 to 186), and care from an anesthesiologist who does not specialize in obstetrics (odds ratio 4.6, 95% confidence interval 1.8 to 11.5). Those odds ratios sound enormous, and the one for urgency has a very wide range around it, which means it is imprecise. Put beside the absolute figure they belong with: across all cesareans with an epidural already in place, 4.3% ended in a general anesthetic.
"Is there an anesthesiologist in the building right now, or on call from home?"
"If it does not work properly, what is the plan, and how long does that take?"