2:43:24 at the 2023 Berlin Marathon, age 45. The result came with a number I didn't think about much at the time. 78.73%, age-graded. AG place 82, out of more than 45,000 finishers.
Out of how many, though, and what would that number have meant if it read 58.73% instead? That's the question most people actually have. A percentage means nothing until you know the shape of the field behind it.
So I graded the whole field. Every Berlin finisher from 2021, 2022, and 2023, all 100,807 of them, run through the same Alan Jones 2025 road tables RaceRecords uses, then sorted by age band and sex.
Here's what my 78.73% actually bought: top 5% of men my age, comfortably. And here's what a 64% buys, or a 52%, because that's where most of the field lives.
What age-grade % does the average finisher score?
Here's the first surprise, if you've absorbed the usual age-grade folklore. The line you always hear, "70% is a good club runner", describes a small minority of a Major marathon field. (I broke down what age-graded percentage actually means separately.)
Most finishers land between 45% and 65%. The modal band is 50–55%, holding 19.3% of everyone on its own. Stack the buckets up and the picture is a hill with a long, thin tail to the right:
- Under 60%: about 72% of the field.
- 60–70%: 20.2%.
- 70% or higher: 7.9%, roughly 1 in 13.
- 75% or higher: 3.6%, roughly 1 in 28.
- 80% or higher: 1.2%, roughly 1 in 83.
That last number is the one I keep coming back to. Eighty percent gets thrown around as a benchmark a committed amateur can chase. In a 35,000-strong Berlin field it's about four hundred people. My own 78.73% didn't quite reach it, and Berlin is one of the fastest fields on earth.
But the median tells the kinder story. Half the field grades under 53%, half over. If your number starts with a 6, you already ran better than three out of four people who crossed the same line.
Does age-grade % rise with age?
Easier or harder with age? Neither, really. The shape barely moves.
Each bar is one men's age band, totalling 100% of the finishers in it. What strikes me is how stable the mix is. The 70%+ slices, the orange ones, stay thin from the twenties all the way into the late sixties. The 80%+ sliver never clears 2.5% in any band.
The 60–70% chunk does swell a little with age, from around 18% in the twenties to 26% in the late fifties. That isn't older runners getting faster. The tables credit the same finish time a little more each year, and the casual end of the field thins out as the age bands climb. The people still racing at 58 tend to be the ones who never stopped training.
Pick your own band and the read is simple. Under 50% is the back half. Anything in the orange and you're in the front carriage.
What age-grade % puts you in the top 5%?
If you want a target instead of a description, here's the table I'd have wanted before my own race. The median is the middle of your band. Top 5% is the 95th percentile, the number you need to beat to be one in twenty.
| Age band | Median (M) | Top 5% (M) | Median (F) | Top 5% (F) |
|---|---|---|---|---|
| 20–29 | 52.3% | 74.6% | 51.1% | 68.1% |
| 30–34 | 53.8% | 74.8% | 50.8% | 69.2% |
| 35–39 | 53.0% | 73.2% | 50.8% | 67.9% |
| 40–44 | 54.1% | 73.5% | 50.9% | 67.1% |
| 45–49 | 55.5% | 75.4% | 52.3% | 68.8% |
| 50–54 | 55.8% | 74.7% | 53.4% | 69.3% |
| 55–59 | 55.9% | 73.4% | 55.0% | 70.1% |
| 60–64 | 56.4% | 74.7% | 56.8% | 73.4% |
| 65–69 | 55.9% | 73.3% | 59.3% | 75.4% |
Look down the Top 5% column for men. It hardly moves, 73 to 75 percent, every band, every decade. To be one of the five fastest men per hundred in your age group at Berlin, you need roughly the same age-graded number at 65 as at 25. I find that quietly reassuring. The bar for "excellent for your age" doesn't drift away as you get older.
Take the 45–49 men's band, since it's mine. The median is 55.5%, the top quarter starts at 63.0%, and the top 5% begins at 75.4%. My 78.73% cleared that line with room to spare, which is the same thing AG place 82 was telling me, said two ways. A 70% in that band, by contrast, is comfortably top 10% but short of the top 5%. (Check your own race against it in the calculator.)
Men, women, and the line that crosses
Now the result I didn't expect. Age grading is meant to be sex-neutral, every runner measured against the world best for their own sex. Equal footing, in theory.
In the Berlin field it isn't equal. Men's median age-graded percentage runs about 2.5 points above women's for most of the age range.
But watch the lines near the right. The men's median is flat, parked around 54 to 56% the whole way. The women's starts lower, near 51%, then climbs. In the early sixties it crosses over and keeps going, hitting 59.3% in the 65–69 band against 55.9% for the men.
I don't read that as women ageing better than men. It's who turns up. A 67-year-old woman at the Berlin start line is a heavily selected athlete in a way a 35-year-old man in the same field is not. The deeper, more casual male field drags the male median down. The thin, committed older-women field pulls theirs up. The maths is fair. The start line is not a random sample.
How I pulled this together
The honest part. This is one race, measured carefully, not the running population.
Big marathon studies usually stop at finish time. RunRepeat, working with World Athletics, analysed about 97% of US marathon results that way in 2025 (RunRepeat, "The State of US Marathons 2025"). Useful, but time by age is not the same as age-graded percentage by age. That distribution is the gap I wanted to close.
The data is a published, name-free compilation of Berlin results, filtered to 2021 through 2023 (moralescastillo/datasets). I dropped rows with no finish time, and the handful with a sex the age-grade tables can't score. That left 100,807 finishers, 68,561 men and 32,246 women.
The full grading method and both table sets live on the methodology page.
One real limitation, and my own race is the cleanest way to show it. The public results record a 5-year competition class, not your exact age. Graded at my real age of 45, my 2:43:24 scores 78.73%. Graded at the band's mid-age of 47, which is what I had to use for everyone in this dataset, it reads 80.14%. A point and a half, on me. That's the size of the error baked into every number in this post. The age factor moves only a little across five years, so the error stays small, but it's real.
So, your number. Over 53% and you beat half of Berlin. Over 63% and you're top quarter for your age. Touch 75% and you're one in twenty, at any age. Mine took twenty years of running to reach, and the clock had already told me what it thought of the race. This is the second opinion.
Sources
- moralescastillo, Berlin Marathon results dataset (1974–2023), 2021–2023 subset — retrieved 2026-05-22, https://github.com/moralescastillo/datasets
- RunRepeat (with World Athletics), "The State of US Marathons 2025" — retrieved 2026-05-22, https://runrepeat.com/the-state-of-us-marathons-2025
- Alan Jones, Age-Grade Tables (road, 2025 edition) — retrieved 2026-05-22, https://github.com/AlanLyttonJones/Age-Grade-Tables
- RaceRecords, Methodology — retrieved 2026-05-22, https://racerecords.run/methodology
- Aggregates and field counts: author's analysis of 100,807 finishers, content-export/berlin-ag-2021-2023.json