How much do hills slow you down? Pace by gradient

The short answer

• Each 1% of uphill costs roughly 5.5% more time per km near flat. At 5:00/km that is 17 seconds per km.
• A 5% climb takes 30.1% longer: 5:00/km becomes 6:30/km for the same effort.
• A 5% descent saves 23.7%: 5:00/km becomes 3:49/km.
• Up and back down never breaks even. Over a hill that climbs and descends at 4%, the round trip takes 2.1% longer than the same distance on the flat. At 2% it is 0.5%.

Pace by gradient table

Read across from a gradient to the pace that matches the effort of your flat pace. “Change” is the extra or saved time against flat ground, the same for every pace.

How to use the numbers

The column you read from is your flat pace at race effort, not your goal pace on a hilly course. On a course that climbs, the flat pace has to be a little quicker than the goal average, because the climbs cost more than the descents give back. Working that out by hand for every kilometre is tedious, which is what the planner does: it finds the one flat pace that makes every graded split add up to your goal time.

Gradient means rise over distance: a 2% grade climbs 20 m in a kilometre. Road marathons use the middle of the table: across the courses we have mapped, the steepest whole kilometre is a 3.7% descent (Boston Marathon, km 1). See where each course climbs on the course pages.

Why a hill costs more than it gives back

Uphill, you lift your body weight on every stride, and that cost grows quickly with the slope. Downhill, gravity helps, but your legs spend energy braking, so the saving grows more slowly. At 4% the climb adds 23.6% and the descent saves only 19.5%. That gap is why a rolling course is slower than a flat one at the same effort, and why a runner who holds even pace over hills pays for it later in the race.

The saving does not grow forever. On this curve the cheapest slope is about 18% downhill; steeper than that, braking costs more again. Long before that point, your quads, the footing and the risk of falling set the limit, so treat the downhill rows as the most you could take, not a target to chase.

Where the numbers come from

The table is the cost-of-running curve that gpx2pace uses for every plan. It is the polynomial from Minetti and colleagues (2002), who measured the energy cost of running on a treadmill at slopes from −45% to +45%. We assume pace scales with energy cost, so an effort that costs 10% more takes 10% longer per kilometre.

It is a model, measured in a lab on trained runners, and it leaves out heat, wind, surface and fatigue. Other curves exist: Strava fitted its own grade-adjusted pace model to heart-rate data from its users, and it differs most on descents. For pacing a road race the differences are a few seconds per kilometre; the shape, slower up and less-than-equally faster down, is the same.