A 21:30 5K Says 3:25, Eight Weeks of Training Say 3:33

Write down three marathon predictions before race week, not one. The first comes from a short race, the second from a half marathon run along the way, and the third from nothing but the training log. When all three agree, the goal pace is easy to choose. When they do not, the size of the gap tells you more about the coming race than any of the three numbers alone.

This article follows one spring marathon block from the first week to the last. The runner is an ordinary amateur: a 5K time trial of 21:30, a tune-up half marathon with a 1:40 goal, and around 30 miles a week. The numbers are round on purpose, so every step can be repeated in the calculators linked along the way, and the three screenshots show the exact inputs.

The short version, before the details:

• Week 1, 5K time trial in 21:30: VDOT 45.8, a marathon near 3:25.
• Week 6, half marathon in 1:40:00: VDOT 45.1, a marathon near 3:28.
• Weeks 4 to 11, 30 miles a week at 8:40 per mile: the Tanda formula says 3:33:14.

Eight minutes separate the fastest and the slowest prediction. The rest of the article is about where those minutes come from and which one to trust on race day.

Week 1: a 5K time trial sets every training pace

The block starts with a solo 5K on a flat loop, run hard but evenly: 21:30, an average of 6:55 per mile. Typed into the VDOT calculator, that time gives VDOT 45.8. VDOT is Jack Daniels' name for an effective VO2max: not a lab value, but the oxygen uptake that would explain this race if running economy were average. Its real use is the list of training paces that comes with it.

The M row is the first marathon prediction: 7:50 per mile over 26.2 miles is about 3:25. A marathon of 3:25:15 gives the same VDOT 45.8, so the two times are equivalent on the Daniels tables.

The same calculator shows how quickly the paces move with the 5K time. Each step in the table below is a minute or two at the 5K, and it shifts the easy range by more than half a minute per mile:

Why the first prediction is the most optimistic

The VDOT tables assume that the runner has trained for the distance they predict. A 5K says a great deal about speed and VO2max and almost nothing about the last 6 miles of a marathon, which depend on fuel, muscle damage and how many long runs are in the legs. That is why a prediction from a 5K is the best case. It is the right number for setting training paces, and the wrong one to copy straight onto a pace band.

Week 6: a half marathon as a dress rehearsal

Halfway through the block comes a tune-up half marathon. On VDOT 45.8 the equivalent half is about 1:38:45, but a tune-up is not a goal race, so the target is a rounder and slightly safer 1:40. The question for the half marathon splits calculator is how to spread those 100 minutes over 13.1 miles.

The runner chooses a negative split of 10 seconds per mile: 7:43 per mile for the first half and 7:33 for the second. On paper that means passing halfway at 50:33, 33 seconds behind even pace, and running the second half in 49:27, 1:06 faster than the first. The split table under the result is the part worth printing. At mile 6 the watch should say 46:16, which is 30 seconds behind an even 7:38. Being behind at that point is the plan, not a problem.

The third row is the quiet favorite of many coaches. The first 5K at 7:49 per mile puts the runner 36 seconds behind even pace at the 5K mat, at 24:18, and the remaining 10 miles at 7:34 per mile make it back by the finish. It costs little in the first mile, where the crowd usually holds everyone back anyway.

Goal pace for other half marathon times

The same calculator opens with the goal already filled in for the most searched half marathon times. Each link below shows even-pace splits for every mile, ready to switch to a negative split:

What the half marathon says about the marathon

The runner finishes the tune-up in exactly 1:40:00, close to the negative split plan. Back in the VDOT calculator, a 1:40:00 half gives VDOT 45.1, a little below the 45.8 from the 5K, and a marathon pace of 7:56 per mile. That is the second prediction: about 3:28, since a 3:28:00 marathon also gives VDOT 45.1.

The drop from 45.8 to 45.1 is typical. A half marathon takes more endurance than a 5K, so the runner whose training has more speed than volume always looks a little weaker over the longer race. Here the gap is 0.7 VDOT points, about three minutes on the marathon. It is a first hint of what the third prediction will show more clearly.

Weeks 4 to 11: eight weeks of training logs

The third prediction ignores races altogether. Giovanni Tanda built it from 46 marathons of 22 amateur runners: the marathon pace follows from two numbers, the average weekly distance and the average pace of every run, warm-ups and recovery jogs included, over the eight weeks that end a week before the race. In this block those numbers are 30 miles a week and 8:40 per mile.

The Tanda marathon predictor answers 3:33:14, a pace of 8:08 per mile, and for about two runners in three in the study the real result fell between 3:29:10 and 3:37:17. Entering the totals from a running app gives exactly the same answer: 240 miles in 34 hours 40 minutes over the eight weeks is the same 30 miles a week at the same 8:40.

Both inputs fit inside the study's data, 25.1-68.8 miles a week and 6:48-8:52 per mile, so this is not an extrapolation. It is also the most honest prediction of the three, because it is built only from what the runner has actually done.

Three predictions, one runner

The three numbers are not in conflict. They answer three different questions. The 5K says what the engine could do with marathon-specific training behind it. The half says what it can do over a longer distance today. The training log says what this particular amount of running usually produces over 26.2 miles. When the third is slower than the first two, the missing piece is almost always volume, and the calculator even prints the warning: below 31 miles a week the biggest risk is not the pace but the last 6 miles.

What closes the gap: miles or pace?

The Tanda formula has two levers, and the predictor shows both under the result. The table below keeps one lever fixed and moves the other:

On paper, running every mile 30 seconds faster closes almost the same gap as adding 10 to 15 miles a week. In practice it is not the same choice. The average of 8:40 already sits faster than the easy range from the 5K, 8:50-9:44, because it includes the threshold and interval days. Pushing it to 8:10 would mean running most easy miles near threshold pace, which is the classic way to arrive at race week tired or injured.

Adding miles at easy pace is the lever the formula rewards most safely. At 45 miles a week and the same 8:40 average, the training prediction becomes 3:24:04, which is where the 5K said the runner could be. The two predictions meet when the training catches up with the speed.

Five lessons from one marathon block

1. Use the short race for paces, not for the goal. The 5K gave the most useful number of the block, the easy, threshold and interval paces, and the least reliable marathon time.
2. Plan the tune-up half as a pacing test. A negative split of 10 s per mile costs only 33 seconds at halfway and shows whether the last 4 miles can hold the pace.
3. Treat a falling VDOT as a message about endurance. 45.8 from the 5K and 45.1 from the half is a small drop, but it points the same way as the training log.
4. Trust the training log most when it is the slowest. A prediction built from what was run is harder to fool than a prediction built from a good day.
5. Close a gap with easy miles, not with faster easy runs. Ten more miles a week are worth about 6 minutes; the same minutes from pace would cost the easy days.

A race-day goal for this block

With 30 miles a week, a sensible goal sits between the training prediction and the half marathon figure: 3:30, or about 8:01 per mile, with halfway at 1:45:00. That is slightly faster than 3:33:14 but well inside the study's range for this training, and 2 minutes slower than the half marathon suggests. If the long runs of 19-22 miles in the last month go well, the goal can move toward 3:28. If they do not, 3:33 is the honest target, and the weather pace adjustment calculator will say how much a warm morning adds on top.

These calculators, the 10 goal pages for the half marathon and the training tools below are enough to repeat this whole block with your own numbers: one time trial, one tune-up race, eight weeks of logs.

Tools discussed in this article

• VDOT Calculator
• Half Marathon Splits Calculator
• Tanda Marathon Predictor

More sports and running tools

• Running Pace Calculator: pace, time and distance for any run
• Weather Pace Adjustment Calculator: how much heat, humidity and wind slow a goal pace
• Interval Training Calculator: repeats and recoveries for a track session
• VO2max Calculator: aerobic capacity from a field test
• Training Heart Rate Calculator: heart rate zones for easy and hard days
• Resting Heart Rate Calculator: what a morning pulse says about fitness
• Walk-Run Calories Calculator: energy used per mile walked or run
• Calories Burned Calculator: calories for a workout by activity and time