Should you run a negative split? What pacing data says
Even or near-even pacing is the safer plan for long races, and a fast start raises the risk of a late collapse. Here is how to set your first half.
By Runapt
Not always. For long races such as a half marathon or a marathon, an even or near-even pace is the better default, and a negative split works only when it comes from a patient start, not from a late surge (Abbiss & Laursen, 2008).
A negative split means the second half is faster than the first. As a slogan it is fine. As a rule it hides the part that matters, which is the first half.
The advice sounds right because the first kilometres lie
The first kilometres of a race feel easy. You are rested, the crowd is loud and the adrenaline is up. A runner aiming for a 4:00:00 marathon needs 5:41/km (9:09/mile). In the first 5 km, 5:25/km feels like jogging.
That gap is where "start slow, finish strong" comes from. The slogan is a defence against a real mistake, going out too fast. The trouble is that it swaps one target for another. "Run the second half faster" gives you a finish to chase, but it says nothing about how fast the opening should be.
So the useful question is not whether the second half is faster. It is how far apart the two halves are, and which half takes the risk.
What the sources measured, and what they did not
Abbiss and Laursen (2008) wrote a review of pacing profiles in athletic competition. They describe negative, even, positive, all-out, parabolic and variable pacing. For prolonged events, they conclude that an even or near-even pace tends to be optimal. That is the claim, and it is a claim about the whole shape of the race, not about a single split.
Smyth (2021) looked at the other end of the problem: a large dataset of recreational marathon runners. The paper shows that a late-race pacing collapse, the thing runners call hitting the wall, is common and costly. It is also more likely after a fast start.
Neither source says a negative split is wrong. Neither says it is a target to chase. What they support is narrower: a pace that holds is better than a pace that starts high and falls apart, and fast starts raise the odds of the fall.
Where the advice breaks down
Take a hypothetical runner with a goal of 4:00:00 for the marathon. Three ways to run it, using plain arithmetic:
Even: 5:41/km all the way. Halfway at 2:00:00.
Fast start: first half at 5:30/km, which is 1:56:02 at halfway. A runner who starts like this feels brilliant. If the second half then slows to 6:10/km, it takes 2:10:06 and the finish is 4:06:08. The nearly four minutes banked early are given back, with six more minutes on top.
Forced negative split: first half at 6:00/km, which is 2:06:35. To still finish in 4:00:00, the second half has to run at about 5:23/km. That is 18 seconds per kilometre faster than the even pace, at the point of the race where legs are heaviest.
The second case is the one Smyth (2021) describes: a fast start followed by a collapse. The third is the one the slogan can produce. Starting deliberately slow in order to run a negative split moves the difficulty into the back half, and you end up asking for a pace you may not have.
Neither runner needed to be unfit. Both just picked a shape that put the risk late.
Wind changes this too. A headwind on the dijk in the first 10 km makes 5:41/km cost more than it does on a calm day. Run that stretch by effort, not by the watch, and expect the pace to come back on the way home. That is a negative split for free, and it is a fine one.
The version worth keeping
Aim for even effort, allow a slightly slower start, and let the second half be as fast as it turns out to be. A planned negative split of a few seconds per kilometre is fine when it comes from that patient start. A 5 to 10 second per kilometre gap is a sensible size, but treat it as one reasonable option, not a law.
Here is a way to set the first half:
Take your goal pace from a recent race or time trial, not from a hoped-for finish time. Runapt sets training paces from a recent race using VDOT, and the same race is the honest starting point for a goal pace.
Run the first 3 km about 5 to 10 seconds per kilometre slower than goal pace. For a 5:41/km goal that is 5:46 to 5:51. Ten seconds over 3 km costs 30 seconds in total, which is easy to recover.
Settle on goal pace at 3 km and hold it to halfway. If the pace feels like work before halfway, the goal is too ambitious, not the plan too slow.
Keep it steady until the last quarter of the race. Only then, if the effort still feels controlled, let the pace drift faster. In a marathon that is roughly after 32 km.
Check step 1 well before race week. A goal pace that does not match recent fitness is the usual reason a start goes wrong. The marathon pace chart shows what each finish time asks for per kilometre, so you can see how big the jump from your current pace is.
Practise this in training, not for the first time on race day. A steady block at goal pace late in a long run, once easy long runs are already a habit, shows what the pace feels like at fatigue. Keep it short. The long run itself stays easy.
If you are unsure the goal fits at all, the free race reality check in Runapt tests a goal time against your calendar, weekly distance and long run. Do it before choosing the pace, since the pace is only as good as the goal behind it.
Sources
Abbiss, C. R., & Laursen, P. B. (2008). Describing and understanding pacing strategies during athletic competition. Sports Medicine, 38(3), 239–252. https://doi.org/10.2165/00007256-200838030-00004
Smyth, B. (2021). How recreational marathon runners hit the wall: A large-scale data analysis of late-race pacing collapse in the marathon. PLOS ONE, 16(5), e0251513. https://doi.org/10.1371/journal.pone.0251513