Garmin Training Readiness: What Actually Goes Into the Score

Your watch gives you a number between 1 and 100 before you’ve had coffee. It says 34, Low, and you have a threshold session on the plan. You slept fine. You feel fine. Do you move the session, or do you ignore the watch and run?

Neither, until you’ve looked underneath the number. Training Readiness is a composite, and composites lie by averaging. A 34 driven by 19 hours of remaining recovery time from Tuesday’s intervals means something completely different from a 34 driven by an HRV status that flipped to Unbalanced four days ago. One is a normal, expected consequence of the plan working. The other is a signal that something in your life outside running has changed and your plan hasn’t noticed.

What is Garmin Training Readiness, mechanically

Garmin Training Readiness is a 1 to 100 score produced by Firstbeat Analytics from six named inputs, refreshed through the day. It appears on the Forerunner 255 and up, Fenix 7 and 8, Epix Pro, Enduro, and Venu 3. The four bands are Low (1 to 24), Moderate (25 to 49), High (50 to 74) and Prime (75 to 100).

The six inputs:

Note what isn’t in that list. Body Battery is not an input, despite sitting next to Readiness on the same glance and moving in roughly the same direction. Neither is resting heart rate, subjective feel, muscle soreness, or anything at all about the mechanical damage from a long run. Garmin has no way to know your quads are wrecked.

Garmin doesn’t publish the weights, and anyone who tells you the exact percentages is guessing. But the behaviour is consistent enough to reverse-engineer in practice, and two inputs do most of the work.

Recovery time is usually the one moving the number

Recovery time is the loudest input and the most deterministic. It’s set by the anaerobic and aerobic Training Effect of your last session, then ticks down in real time, reduced further by good sleep and daytime rest. It caps at 96 hours on most Forerunners.

Here’s the thing that trips people up: recovery time is assigned by physiological strain during the session, not by how the session felt afterwards. A 40-minute session of 8 x 400m at 5K pace with a 4.6 anaerobic Training Effect can hand you 52 hours of recovery time. A 2-hour steady long run at easy pace, which will leave you far more beaten up on the stairs, might only set 26 hours because it never pushed you far into anaerobic territory.

So the morning-after-intervals score in the 30s is almost always recovery time, and it is not a problem. It’s the model telling you a hard session did what a hard session does. If your Runna plan has an easy 6K scheduled for that day, run the easy 6K. The score isn’t vetoing it.

Redrawn as text, the factor list on a typical post-intervals morning looks like this:

TRAINING READINESS            34   LOW
────────────────────────────────────────
Sleep score            81           good
Recovery time          19h     ↓ limiting
HRV status         Balanced          good
Acute load             612        optimal
Sleep history          7h 24m        good
Stress history         33        moderate

One limiting factor, and it’s the one that expires on a timer. Scroll to this list before you make any decision about your training day. On the watch it’s the Training Readiness glance, then down; in Garmin Connect it’s under the score on the Training Readiness card.

HRV status is the one to actually take seriously

When the factor list shows Balanced everywhere except HRV status, the picture changes. HRV status compares your 7-day average overnight HRV against a personal range Garmin builds from three weeks of consistent overnight wear. Say your range is 44 to 71 ms. A 7-day average of 39 ms puts you Unbalanced on the low side, and because it’s a seven-day average, it will not reset overnight. You are looking at four to ten days of suppressed Readiness scores.

Unbalanced-low is the input that should make you change the plan, because it’s the one that tracks the things a training app cannot see: a bad work month, a head cold you haven’t admitted to, three nights of wine, heat acclimatisation, a long-haul flight. It’s also worth knowing what contaminates it. Wrist-worn HRV sampling is sensitive to strap fit, and a watch that’s loose enough to slide down your wrist overnight will produce garbage overnight HRV that a chest strap wouldn’t. Before you rewrite six weeks of a marathon block, check whether you changed watch position or started sleeping in a hotter room.

Two weeks out from a goal race, the taper case catches people out. You cut volume, your acute load drops from 780 to 410, and you expect Prime scores every morning. Instead you get High, occasionally Moderate, because Garmin reads a rapidly falling acute load with some suspicion and because taper often comes with worse sleep, not better. This is normal. Judge the taper by how workouts feel at goal pace, not by the readiness number.

The 20-miler paradox

Run 32 kilometres on Sunday at a genuinely easy effort, sleep nine hours, and Monday’s Training Readiness can easily read 84, Prime. Your legs disagree strongly. Both are correct within their own terms: your autonomic nervous system has recovered, and your connective tissue has not.

This is the single most important limitation of the score for marathon runners, and it’s where an AI-built plan and a Garmin score can reinforce each other into a bad week. The plan doesn’t model soreness. The watch doesn’t model soreness. If you use a Prime score as permission to add a quality session on top of a big weekend, nothing in the loop is going to stop you. Garmin Training Readiness is a nervous-system-and-sleep score wearing the label of a whole-body score, which is a useful distinction to hold onto when you’re reading any of the recovery and readiness metrics your watch produces.

A rule that survives contact with a real training week

Work through the factor list in this order and act on the first thing that’s limiting:

1. Recovery time is the only limiter. Proceed with the session as planned if it’s easy or moderate. If it’s a hard session and recovery time is above roughly 24 hours, you’re stacking quality too close together; that’s a plan structure problem, not a today problem.
2. Sleep score is the only limiter, one night. Run the session. One bad night doesn’t meaningfully blunt a threshold workout, and the data on this is pretty unglamorous.
3. Sleep history is limiting across three days. Downgrade quality to aerobic. Sleep debt does blunt adaptation.
4. HRV status is Unbalanced or Low. Hold volume, cut intensity, for as long as it takes to come back into range. If it’s still Unbalanced after ten days and nothing obvious explains it, that’s a conversation with a GP, not a tweak to your plan.
5. Acute load is flagged high and HRV is drifting. Your AI plan is progressing you faster than you’re absorbing it. Cut the next week’s volume by 20 percent and watch whether HRV status recovers.

Where this gets genuinely useful is in the feedback loop with a generated plan. Runna and Garmin Coach both adjust from completion and pace, not from recovery inputs, and a ChatGPT plan adjusts from nothing at all unless you tell it. You are the missing sensor. Log your readiness factor list, not the score, on the mornings after your two hardest sessions each week, and you’ll see within three weeks whether the progression rate is one your body is actually accepting.

Tomorrow morning, before you look at the number, scroll past it.