Guides · Improve your 5K time
How to improve your 5K time on a treadmill, and know the gain is real
Your turkey trot is on Thanksgiving morning, so you ran a 5K test on the gym treadmill at 6.6, and at that setting a 5K takes 28:15. You want to break 27:30 that morning, and your plan is to retest on a belt in November and see if you are there.
In one six-week study of 20 well-trained runners, 5K times were about 2 percent faster on average after hard uphill intervals. Retest on a belt you have checked. In one lab study, belts on four treadmills, one of them worn, averaged 98 to 103 percent of their setting, which is about as big as that gain.
What actually moves a 5K
In one six-week study, 20 well-trained runners were each put on one of five uphill interval programs. The programs ranged from short sprints up a very steep grade to longer reps up a gentle one. Across all five, 5K times improved by about 2 percent on average, and no grade came out clearly best. The authors concluded that until more data comes in, runners can assume any form of hard uphill interval training will benefit a 5K.
Those runners were already trained, with best 5Ks around 16 and a half minutes, and each program had a small group. The study tells you what happened to them over six weeks, not how much you will gain.
Three guides on this site cover what a treadmill does to each of those sessions. The hill repeats guide shows how to compare hill sessions at different grades and speeds, so a steep day and a gentle day sit side by side. The 800s guide covers what a short, hard rep on a belt can tell you about your speed, and what it cannot. The tempo guide covers the longer steady efforts between those days, and what a belt a few percent off does to them.
That study timed its 5Ks on an outdoor track, a distance nobody had to take on trust. Your retest will run on a belt whose real speed you may never have checked.
Your whole block is one or two presses of the button
The speed behind a 5K, and what 2 percent faster takes
Each row assumes the belt runs the speed shown.
| Goal time | Speed needed (mph) | 2 percent faster | New speed (mph) | Change (tenths) |
|---|---|---|---|---|
| 30:00 | 6.21 | 29:24 | 6.34 | 1.3 |
| 27:30 | 6.78 | 26:57 | 6.92 | 1.4 |
| 25:00 | 7.46 | 24:30 | 7.61 | 1.5 |
| 22:30 | 8.28 | 22:03 | 8.45 | 1.7 |
| 20:00 | 9.32 | 19:36 | 9.51 | 1.9 |
If your block brings a gain like that study's average, it shows up on the treadmill as one or two more presses of the speed button that you can now hold. That is the whole signal you trained for.
The belt can be off by the same one or two presses
Pressing 6.6 tells the machine what you want. The display shows that request, and the belt does not always deliver it exactly. Researchers measured the belt speeds of four lab treadmills as people walked and ran, and one machine was heavily worn. Across the speeds and machines tested, average belt speeds landed from 98 percent of the setting up to 103. If a belt at 6.6 sat anywhere in that range, it would run from about 6.47 to 6.80, a press or two either way.
Think of topping up your bike tires. You want 2 more psi. You check the pressure at home with your own gauge, then pump at the shop with theirs, and the two gauges disagree by about 2 psi. The change you were after is the same size as the gap between the gauges, so you cannot say whether you added any air at all.
A retest on a different treadmill has the same problem.
Two tests, two belts, one wrong conclusion
These are made-up numbers, and both belts sit inside the lab range. The settings say you got 50 seconds faster, while the true times say you got 17 seconds slower.
| Test | Setting (mph) | Time at the setting | If the belt ran | True time |
|---|---|---|---|---|
| October, gym treadmill | 6.6 | 28:15 | 2 percent fast | 27:41 |
| November, home treadmill | 6.8 | 27:25 | 2 percent slow | 27:58 |
You would line up on Thanksgiving chasing 27:30 on a gain you never made, and the first mile of the measured course is where you would find out. A slow belt first and a fast belt second can make a small real gain show as a slower time. A block that worked then looks like one that failed.
Make the retest a real one
Use one treadmill for both tests, because that takes the second gauge out of the comparison. The guide to repeats on a treadmill explains why a session on a different treadmill cannot separate your fitness from that belt's speed.
One machine still leaves two questions. Did that belt hold the same speed on both days, and what real speed is your gain sitting on? Your watch's pace may be no help here. Two major watch makers say that on indoor runs their watches use a motion sensor inside the watch. That sensor sits on your wrist, not on the belt, and the watch guide explains where that pace comes from. The race on a measured course gives the final word. If your belt proves far off its setting, that is a matter for a technician or the owner's manual.
Know the belt behind both numbers
Measure the belt at your test setting before each test. Then your before and after rest on measured speeds, not the number on the display. Treadwell takes that measurement with an iPhone Pro's camera and LiDAR, pointed at the belt.
Check the belt before your retest
Get a measured speed under your 5K test, then do it again in November. You will need an iPhone Pro for Treadwell.
Scan with your iPhone Pro to get the appTested: in our tachometer tests the app read within about 2 percent of true belt speed.
Quick answers
How do you improve your 5K time on a treadmill? In one six-week study of 20 well-trained runners, 5K times were about 2 percent faster on average after hard uphill intervals. Retest on a belt you have checked. In one lab study, belts on four treadmills, one of them worn, averaged 98 to 103 percent of their setting, which is about as big as that gain.
What treadmill speed runs a 5K in 30 minutes? About 6.21 mph for 30:00, and about 7.46 mph for a 25:00 5K. Both numbers assume the belt runs the speed shown.
What treadmill workouts improve a 5K? Hard uphill intervals have a study behind them. In one six-week study of 20 well-trained runners, five uphill interval programs, from short sprints up a steep grade to longer reps up a gentle one, improved 5K time by about 2 percent on average, and no grade came out clearly best. The hill repeats and 800s guides on this site cover what a belt does to those sessions.
Is a treadmill 5K faster than outside? Not reliably. A 2019 review of 34 crossover studies found endurance performance, mostly time-trial duration, was poorer on a motorized treadmill than over ground, on average. That is a group result with no set number of seconds, so it cannot convert your own time from one to the other.
Sources.
- Uphill intervals and the 5K: Barnes K.R., Hopkins W.G., McGuigan M.R. and Kilding A.E., Int J Sports Physiol Perform 8(6):639-647 (2013), PMID 23538293. Twenty well-trained runners did one of five uphill interval programs for six weeks; mean 5K time-trial improvement across all programs was 2.0 percent (limits plus or minus 0.6), with no clear optimum grade. The time trials took place on a 400 m outdoor track.
- Belts against their settings: Willwacher et al. (2021), published in Scientific Reports, volume 11, article 2244 (PMC7838258). The team measured the belt speeds of four laboratory treadmills, one whose belt and deck were substantially worn, while participants walked and ran; mean belt speeds sat at about 98 to 103 percent of target speed, a range that belongs to those four machines.
- Treadmill versus overground performance: Miller et al. (2019), Sports Med 2019;49(5):763-82, a meta-analysis (PubMed 30847825). In a meta-analysis of 34 crossover studies, endurance performance (time-trial duration) was poorer on a motorized treadmill than overground, as a pooled average across the studies that measured it.
- Indoor watch data: Apple's support document 105002 describes indoor running as also using the watch's accelerometer, and the Indoor Activities pages of Garmin's Forerunner 165 manual say the accelerometer supplies indoor speed and distance with GPS off.
- 5K = 5,000 m, or 3.1069 miles. The table speeds, the 6.47 to 6.80 span, and the October and November runs are my arithmetic on that distance. The runner, the two belts and the tire gauges are my own illustrations, not logged tests.