One Rough Night, One Rising Week: Reading HRV

Last night's sleep score read 64, about 5% under its 7-day average of 67.4. Total sleep time was 7.2 hours, with resting heart rate at 51 bpm. Overnight HRV told a different story: 89 ms, up 8.5% from its own 7-day average of 82 ms. The day's two main recovery signals pointed in different directions.

Sleep architecture that night shifted too: deep sleep totaled 0.3 hours and REM sleep 0.8 hours, against a Body Battery reading of 95/100 at wake. None of those numbers on their own say whether a planned hard day should hold.

Zoom out past one night and the picture sharpens. The HRV average across the most recent 7 days climbed to 85.9 ms, up 9.7% from the 78.3 ms average of the 7 days before that. A single lower sleep score did not reverse a week of rising HRV.

Why a multi-day average, not one reading

A 2007 study tested this directly in 26 healthy, moderately fit men over 4 weeks. The group whose training was guided by a daily HRV reading used the 10-day average and a short trend, not a single measurement, to decide when to train hard or rest. That group raised VO2peak from 56 to 60 ml/kg/min, while the group following a predefined schedule did not change significantly. The change between the two groups was not itself statistically significant.¹

The HRV-guided group's rule in that study was mechanical, not a feeling: easier training or rest was prescribed whenever HRV fell below a reference value built from the 10-day mean and standard deviation, or showed a falling trend across 2 days. A single lower reading, on its own, did not change that day's plan.

A later study followed 40 recreational runners through an 8-week intensive block. The HRV-guided group trained hard on fewer sessions than the predefined group — 13.2 against 17.7 — yet still improved a 3000-meter time trial by 2.1%, a result the predefined group's 1.1% change did not reach. VO2max improved in both groups, by 3.7% and 5.0%. The authors describe the between-group difference as small.²

A systematic review with meta-analysis of randomized controlled trials in trained endurance athletes found a significantly higher effect on VO2max for HRV-guided training than for control training, though the overall effect across all groups was described as small, and the amateur level and female subgroups showed the better results.³

That same meta-analysis found larger effects in amateur athletes and in women than across the pooled sample as a whole — a reminder that a small effect at the group level can still matter more for some individual profiles than others.³

What this means for a VO2 max day

A power meter and a zone chart make a hard interval day easy to plan on paper. The harder call is whether to start it at all when one metric looks off.

On this day, resting heart rate was 51 bpm and overnight HRV rose rather than fell. The 7-day HRV trend behind both numbers kept climbing. The plan called for 43 minutes of zone 5 work, and it stayed confirmed for the day.

1. Check the 7-day HRV average before deciding to skip a hard day, not just last night's reading.
2. Weigh a lower sleep score against the slower-moving multi-day trend, not in isolation.
3. Note that the planned effort here held at 43 minutes of zone 5 work while the 7-day HRV average stood at 85.9 ms.
4. Expect day-to-day HRV swings even inside a rising week: the 2007 study behind this approach built its decision on a 10-day window and a short trend, not a single day's reading.¹

The lesson is not to ignore a lower sleep score. It is to read it alongside the slower-moving number: the week's HRV trend, not one night's reading alone.

Does a rising HRV average always mean it's fine to go hard?

Not on its own. It is one input, alongside the sleep score and the planned session's purpose. On this day the 7-day HRV average stood at 85.9 ms despite a lower sleep score, and the planned 43-minute zone 5 session stayed confirmed.

What should replace checking only last night's reading?

The 7-day average and its direction, the way the 2007 and 2016 studies cited here set up their own decisions: a 10-day window and a short multi-day trend, not a single night's measurement.¹²