Race Time Improvement Explained: How to Measure a PB, and What Counts as Good

Why race time improvement needs two numbers, not one

Every runner knows the moment. You cross the finish line, stab the watch, and the first thing you want is the comparison — how much faster is this than last time? The obvious answer is subtraction: new time minus old time. The race time improvement calculator does that in one field, but it also returns a second number that matters more the further you get from your first race: the percentage change.

Seconds are what you brag about at the pub. Percentages are what actually tell you whether your training is working. Ten seconds off a 20:00 5K is nearly a 1% improvement — hard to come by for anyone under 25 minutes. Ten seconds off a 4:00:00 marathon is 0.07% — statistical noise, well within the range of a pacing decision or a bathroom stop. This guide walks through both numbers, what a realistic gain looks like at every experience level, and the traps that make a PB comparison mean nothing.

The two numbers the calculator returns

The engine behind the race time improvement calculator is arithmetic, not machine learning. Both race times are converted to total seconds:

prev_total_sec = prev_h × 3600 + prev_m × 60 + prev_s
new_total_sec  = new_h  × 3600 + new_m  × 60 + new_s
seconds_saved  = prev_total_sec − new_total_sec
percent_change = seconds_saved ÷ prev_total_sec × 100

A positive seconds_saved means the new time is faster. A negative value means the new time is slower — the calculator flips the label to "slower" and shows the magnitude, so you always see a positive display next to a clear direction. Percentage change uses the standard relative-change formula (identical to how GPS watch vendors like Garmin, Coros and Polar compute PB deltas in their training summaries), and it is the number to log if you want to compare progress across a long time frame.

Worked example: 5K PB comparison

Say your parkrun PB was 25:00 flat two years ago and you just ran 22:30. Feed both times into the race time improvement calculator and the maths goes like this. The previous time converts to 25 × 60 = 1500 seconds; the new time converts to 22 × 60 + 30 = 1350 seconds. Seconds saved = 1500 − 1350 = 150, which the calculator reformats back to 02:30 for readability. Percentage improvement = 150 ÷ 1500 × 100 = 10.00%.

A 10% 5K PB is substantial. It is roughly the difference between someone with a year of consistent training and someone with three or four years of it. At the same relative gain, a 22-minute 5K runner would drop to 19:48, and a 30-minute 5K runner would drop to 27:00. This is why percentage matters — the absolute time drop is very different, but the underlying fitness change is the same.

For a marathon example, take a 4:00:00 finisher who runs 3:45:00. Total seconds: 14 400 → 13 500, saving 900 seconds (15 minutes), which is 900 ÷ 14 400 × 100 = 6.25% improvement. Fifteen minutes off a marathon feels huge; as a percentage it is closer to the gain from a season of consistent 40-mile weeks, not a transformation.

What counts as a "good" improvement

The honest answer is: it depends on how long you have been running. There is no single benchmark, but the ranges below cover most runners at each stage.

First-year runner

Expect 15 to 25% off your starting time in the first year at any distance, especially if you go from no running to a structured plan. This is base-fitness compounding: mitochondrial density, capillary density, and running economy all improve quickly from a low starting point. A first-year runner going from a 32-minute 5K to a 26-minute 5K has improved 18.75% and that is fairly typical rather than exceptional.

Intermediate (2-5 years)

Two to five percent per year at the same distance, on the same course, in similar conditions. Anything above that usually means one of three things: a big block of distance-specific work (marathon-specific long runs, threshold-focused 5K training), coming back from injury or detraining, or the previous race was a bad day. If you are consistently posting 5%+ improvements year over year as an intermediate, that is real progress but the growth curve will flatten.

Advanced (5+ years, sub-elite)

One to two percent per year is a solid year. Many advanced runners plateau for two or three years and then break through when they change coaches, distances, or add a big block of altitude / high-volume work. A sub-3 marathoner moving to 2:55 has improved 2.8% and probably worked harder for that than the first-year runner did to go from a 32-minute to a 26-minute 5K.

Elite

Well under 1% per year on average, with occasional bigger jumps at career milestones. Eliud Kipchoge went from a marathon debut of 2:05:30 in April 2013 to a world-record 2:01:09 in September 2022, an improvement of 3.5% across nine years, or roughly 0.4% per year. At this level a percentage improvement of 1% in a single race is career-defining.

What affects race time improvement

Course and elevation

A hilly marathon vs. a flat one costs 3-8% of finish time for the same fitness. If your previous PB was on a pancake-flat certified course (Berlin, Chicago, Valencia, Manchester) and your new time is on a rolling course (New York, San Francisco, Snowdonia), a 2% slowdown may actually be a fitness improvement. The race time improvement calculator does raw subtraction; it does not know about elevation profile. Log the course alongside the time.

Weather

Warm and humid conditions slow marathon times by 2-4% versus ideal (5-12 °C, 40-60% humidity). Wind, especially at exposed coastal or open-course marathons, adds another 1-2%. Track and cross-country studies from the RunRepeat and Strava datasets consistently show these effects at both elite and mid-pack level.

Pacing strategy

Even pacing — running the second half of a race within 1% of the first — typically produces a 2-5% better finish than a positive split (going out fast, dying at the end). Almost every marathon world record in the last decade has been run with a very slight negative or near-even split. A big positive-split PB attempt often leaves 3-4 minutes on the road at marathon distance for a runner with the fitness to have run a flat pace.

Training-block context

A slower race in the middle of a heavy training block is not a fitness regression — it is fatigue. A tune-up 10K run in week 8 of a 12-week marathon block will often be 30 seconds to 2 minutes slower than a fresh, fully-tapered 10K, even though the runner is fitter. The right time to judge a fitness change is a fully-tapered race against another fully-tapered race on similar courses in similar weather.

Sleep, illness, life stress

Under-slept runners (less than 6 hours the night before) typically underperform by 1-3%. Racing within two weeks of a cold or flu can cost 2-6%. These effects do not average out — if the previous race was on a well-slept, healthy week and the new race was on a red-eye flight and a head cold, the "improvement" number is measuring life conditions, not fitness.

How to log race time improvement usefully

• Log same-distance comparisons first. The race time improvement calculator does raw subtraction, so it is only meaningful over the same distance. Track 5K, 10K, half marathon and marathon PBs separately.
• Note the course and conditions. Write the course type (flat / rolling / hilly) and weather next to the time. Otherwise, in six months, you will not know why the number moved.
• Compare tapered races to tapered races. A tune-up race mid-block should not be compared to a season-goal race. They are measuring different things.
• Use age-graded scores as you age. Once you are past 40, raw PB deltas start to trend the wrong way even when your fitness improves. World Masters Athletics tables convert your finish time into a percentage of the age-and-gender world record — a rising age-graded score is real progress even if seconds are not moving.
• Project between distances with a formula, not subtraction. If you want to compare a 10K to a half marathon, use the Marathon Time Predictor to project one time to the other distance first.
• Log negative improvements too. A season of tune-up times often forms a shallow U — slower in the fatigue phase, faster in the taper. The signed value from the calculator makes this visible.

Common mistakes

Comparing across distances with raw subtraction. A 5K PB and a marathon PB are not commensurable. Runners will sometimes claim "I took 15 minutes off my marathon this year and only 30 seconds off my 5K" as if the two numbers are on the same scale. They are not — the percentages tell a different story, and either the marathon jump was course-and-conditions dependent or the 5K number was already close to a physiological ceiling.

Ignoring course profile and weather. An "improvement" from a hilly, hot marathon to a flat, cool marathon is partly the courses. Anywhere between 2 and 6% of a marathon delta can be explained by non-fitness factors, and refusing to acknowledge that tends to lead to overtraining in the next cycle when the runner tries to reproduce the gain.

Chasing PBs on the wrong course. If you run the same hilly local half every year and your times are stuck, one flat, well-organised half in a different town may show you a 2-4% "improvement" that is really just course. Log the course; do not blame yourself for a plateau that is topography.

Overreacting to a single slow race. One race is noise. A slower time can come from a red-eye flight, a bad night's sleep, a hot day, a tight taper, a stomach bug, or simply a bad day. The pattern across three or four races in a season is signal.

Projecting between distances

The race time improvement calculator subtracts. For cross-distance comparison, use the Riegel formula:

T₂ = T₁ × (D₂ ÷ D₁)^1.06

The exponent 1.06 was derived by Peter Riegel in the 1977 Journal of the American Medical Athletic Association paper "Time Predicting" from a large dataset of amateur race performances. It is now the standard tool for projecting between distances between the 5K and marathon. For example, a 22:30 5K projects to a marathon of 22.5 × (42.2 ÷ 5)^1.06 ≈ 214 minutes, or about 3:34:00 — assuming marathon-specific training. Runners who do not do the specific long-run work often miss the projection by several percent. The Marathon Time Predictor Calculator applies this formula and is the right tool when the two races you want to compare are not the same distance.

When PB deltas stop being the right metric

Past 40, raw PB comparisons start to lose meaning. World Athletics and World Masters Athletics publish age-graded tables that convert a finish time into a percentage of the age-and-gender world record for that distance. An 80% age-graded score is regional-championship level; 90%+ is national-championship level; 100% is a world record. A 50-year-old going from 78% age-graded to 81% age-graded is improving meaningfully even if the raw finish time is slightly slower. If your PB delta from the race time improvement calculator is trending the wrong way but your age-graded percentage is climbing, the fitness is going up — the calendar is just going up faster.

Similarly, if you are training for a specific event — cross-country, ultramarathon, trail — PB deltas on road times can be misleading. Marathon fitness and 100 km trail fitness overlap, but not completely. A trail runner whose road marathon time slipped by 2% while their trail ultra times dropped by 8% has clearly moved fitness in the direction their sport rewards.

Related calculators

The running toolkit on Calc Dragon sits alongside the race time improvement calculator. The Running Pace Calculator works out pace, speed and projected finish times for any distance from a 400 m track rep to a marathon. The Marathon Time Predictor applies the Riegel formula to project between distances when raw subtraction stops being meaningful. The Marathon Pace Calculator gives you the even-effort splits for a goal finish time so you avoid the pacing mistake that costs 2-5% of the finish. The Pace Converter swaps between minutes per kilometre and minutes per mile when you are comparing training with runners on the other side of the Atlantic. And the Calories Burned Calculator estimates energy expenditure by activity, time and body weight so a race-day fuelling plan can be built on something more precise than "eat pasta".