Pacing decisions
When to adjust marathon pace mid-race
A useful adjustment fits the race you are running now. Reassess effort, conditions and the remaining distance before deciding whether the original finish target still makes sense.
Decide whether this is a pacing problem
An unexpected hill, headwind or growing fatigue may justify easing and revising your plan. Chest pain, fainting, confusion or severe breathing difficulty call for stopping and urgent assistance, not another checkpoint test. This guide addresses ordinary pacing decisions when you are otherwise well.
Use four inputs together
- Effort: compare breathing and perceived exertion with what you rehearsed.
- Conditions: assess the actual terrain, heat and exposure.
- Heart rate: consider a credible trend in your own context, not a universal cap.
- Checkpoints: compare the time already taken and the task remaining.
The conditions calculator offers scenarios, not a physiological safety assessment. This guide covers unexpectedly high heart rate.
At 10 km: settle before accelerating
If effort matches the plan, continue without treating a comfortable start as permission to speed up. If effort is substantially harder than expected, ease and reassess on a suitable section. Avoid chasing a crowded opening with a surge; use the first-10K guide to interpret it.
At halfway: consider a new target
Review your breathing, legs, fueling tolerance and what lies ahead. If those checks no longer support the original target, update it. You are not obliged to make the second half faster simply because the first half took longer than expected.
At 30 km: make the remaining pace visible
At 30 km in 2:55, a four-hour finish requires the last 12.195 km at about 5:19.8/km. A 4:10 finish would allow about 6:09.0/km. These are calculated scenarios, not predictions of what you can sustain. The four-hour guide and calculator provide reference splits.
Use an adjustment you can assess
On a small screen, swipe the table to see all columns.
| Situation | Immediate response | Recheck |
|---|---|---|
| Unexpected climb or headwind | Ease effort; allow a slower split | After the exposed section, without a catch-up surge |
| Effort too high on comparable terrain | Reduce pace enough to feel more controlled | The next suitable checkpoint and the effort trend |
| Fuel becoming difficult to take | Ease and use the plan you have practised | Whether symptoms settle; seek aid if they persist or worsen |
| Target requires faster running while you are fading | Revise the target | A manageable next segment rather than the original finish time |
There is no universal 5-, 10- or 15-second correction that resolves every case. Conditions and symptoms determine whether continuing is appropriate.
Avoid automatic catch-up rules
Time lost on a hill is not necessarily available on the descent. Missing a gel does not justify forcing extra fuel you have never tolerated. An elevated heart rate does not establish dehydration. Combine the evidence and avoid abrupt compensations.
My Sydney review records the transition to repeated walking around 31–32 km. My half-marathon comparison records a different late-race pattern in 2026. Neither account proves a universal rescue strategy.
Mid-race questions
Should I keep the original goal if I am behind?
Only if the pace required for the remaining distance is still realistic in the current conditions. Calculate the requirement before treating the target as fixed.
Do I need to wait until halfway to adjust?
No. Respond to a clear mismatch earlier. Concerning symptoms require immediate attention.
Does adjusting mean I chose the wrong goal?
Not necessarily. Conditions and execution can change. A useful plan includes a way to revise expectations.
Sources and limits
- Smyth (2018), Fast starters and slow finishers
- Coyle and González-Alonso (2001), Cardiovascular drift during prolonged exercise
- Burke and colleagues (2019), Contemporary nutrition strategies for distance runners and race walkers
Worked examples are calculations for the situations described. Adapt them to your own training.