This research piece is not as rigorous or polished as usual. I wrote it quickly in a stream-of-consciousness style, which means it’s more reflective of my actual reasoning process.

My understanding of HIIT (high-intensity interval training) as of a week ago:

1. VO2max is the best fitness indicator for predicting health and longevity.
2. HIIT, especially long-duration intervals (4+ minutes), is the best way to improve VO2max.
3. Intervals should be done at the maximum sustainable intensity.

I now believe those are all probably wrong.

I think I got the wrong idea because a lot of HIIT/VO2max promoters cite scientific studies, which makes them seem superficially reasonable, but the studies they cite aren’t very good, or aren’t interpreted correctly.

A few months ago I started incorporating some HIIT into my cardio routine. But I didn’t really know the best way to do it, so last week I decided to do some research. I looked up the most-cited meta-analyses on Google Scholar and I noticed my confusion when the meta-analyses didn’t seem to support the conventional wisdom that HIIT is the best way to improve VO2max.

The most comprehensive single source I found was a meta-meta-analysis by Crowley et al. (2022)¹, which reviewed the findings of meta-analyses on HIT (high-intensity training) vs LIT (low-intensity training) for VO2max. The key quote:

Evidence from the meta-analyses that directly compared LIT versus HIT protocols on VO2max was, ostensibly, reported as either trivial or inconclusive. Three out of the six included meta-analyses reported small/moderate beneficial effects of HIT over LIT (α < 0.05). However, two of these reviews reported “substantial” heterogeneity (I2>0.75), small-study bias (p < 0.10), a relatively small pooled sample size (i.e., <1,000 participants), had a high degree of overlap (CCA = 11%) and reported several moderators (e.g., baseline fitness levels, age, HIT variables [e.g., volume, frequency, and duration]), which likely affected results.

Also, in my naiveté I had assumed that these were meta-analyses of RCTs, but in fact most of the included studies weren’t even RCTs:

Scribbans et al. reported that none of their included studies applied RCTs, Sloth et al. reported only four studies that applied RCTs design, and Gist et al. reported that the majority of included studies were RCTs.

(Note: Gist et al.², which did look mainly at RCTs, found that sprint interval training did not work better than endurance training (Cohen’s d = 0.04, 95% CI = -0.17 to 0.24.)

So I thought, okay, these meta-analyses don’t seem to favor HIIT much if at all. But maybe they’re done by stuffy academics who don’t know anything about real training. Are these meta-analyses considered respectable? So I went to see what the Barbell Medicine³ guys thought. I have a lot of respect for them when it comes to strength training. I don’t know if they know about cardio, but one of them is a former competitive swimmer so probably they know something. And they have good epistemics on strength training, and good epistemics might generalize. They did a podcast⁴ on HIIT with some useful content:

• They started the podcast by criticizing a tweet⁵ in which fitness influencer Rhonda Patrick recommended a collection of “evidence-based HIIT protocols”. Their criticism mainly focused on how (they claimed) the provided HIIT protocols were way too hard. They quoted two responses⁶⁷ by exercise physiologists arguing the same.
• They went on to talk about some of the research on HIIT, citing the same meta-analyses that I’d looked at.
• Their ultimate recommendation (also given in an article⁸): “it is reasonable for about 80% of training to be of moderate intensity (zones 1-2), and about 20% reserved for higher intensity work (HIIT or SIT)”. They also said it’s good to use a variety of HIIT protocols and that there is no single optimal protocol.

I didn’t know if the Barbell Medicine guys were right about any of that, but it gave me some direction.

I had a look at the website of one of the people from that Twitter thread, Steve Magness⁹. He ran a 4:01 mile in high school so he probably has some idea of what he’s talking about.

Now, when some science-literate fitness influencers like Peter Attia and Rhonda Patrick give some recommendations about HIIT, and some other people like Steve Magness and Barbell Medicine disagree with them, I don’t have sufficient expertise to say who’s right. Both sides have the trappings of scientific credibility (e.g. citing multiple studies). But one thing I can do is check their logic.

So I checked Steve Magness’s logic. He wrote in a Twitter thread:¹⁰

You need all intensities to max VO2max. So it’s dumb to pit one vs. other

But research shows continuous likely matches HIIT for Vo2max increase

HIIT appears better when you constrain to 8 weeks but when you look over longer time it equalizes

Here’s data from a recent review.¹¹

You can see HIIT appears to increase VO2max more because of the time frame of most training studies (6-8 weeks).

Intense work gets big boost, then levels off. Endurance work gives Longer more gradual boost.

The referenced review does seem to support Magness’s argument, but I don’t know if the review is any good. What I do know is that Magness’s logic makes sense. It stands to reason that a more intense exercise protocol will cause faster short-term gains, but it can’t keep producing those rapid gains forever. And it stands to reason that if most studies on HIIT vs. LIT only last 6-12 weeks, then they will underestimate the long-term benefits of LIT.

That makes logical sense to me, which makes me think Steve Magness knows what he’s talking about.

He also wrote an article¹² arguing that some people care too much about VO2max for longevity:

Vo2max matters. But it’s just one component of many that make up both performance and aerobic fitness. And that’s important because if we return to the original claims that Vo2max is the key indicator of longevity, we’ll find that the majority of the studies cited did NOT even use Vo2max as the main variable. They used performance! In the majority of research, peak speed and incline during the exhausting test was the main correlate to longevity.

The large study¹³ on 750,000 veterans that found a 4-fold higher mortality risk for low versus high fitness used peak speed and incline, not Vo2max. Same with the research¹⁴ on 120,000 individuals finding a 5x difference in the risk of early death.

That makes logical sense to me. VO2max is only one aspect of fitness (albeit an important one), and it stands to reason that your actual ability to perform physical tasks is a better measure of physical health.

I did also look at some of the evidence cited in that article, specifically the Harber et al. (2017)¹⁵ meta-analysis, and Table 2 confirms Magness’s claim—most studies measured speed, time, or total work performed, not VO2max directly.

Insofar as I can verify Steve Magness’s claims, they seem to be correct. He also claims that HIIT protocols should not be “all-out”—for example, 60-second running intervals should be done between a 5K and a one-mile pace.¹⁶ He doesn’t cite any research on that claim, and as far as I know, there isn’t really research on it, it’s just how most high-performing athletes train. But since he seems right about the verifiable claims he’s made, I expect he’s right about that, too.

On the subject of checking people’s logic, here is an (admittedly cherry-picked) quote from the other side of the argument:

[I]s anything done in a total of 10-minutes that big of a deal?

If anyone is misinterpreting my statement as prescriptive:

My underlying point was that anything you can do in 10-minutes is limited on a relative harm basis, even if you do a lot.

This does not make logical sense to me. You can absolutely hurt yourself in less than 10 minutes. Even putting injury risk aside,

• I’ve done 10-minute hill sprint intervals in the morning that left me feeling tired all day.
• A 10-rep max squat takes less than 60 seconds, but it makes my legs sore for the next two days.

(Maybe that’s less logic and more personal experience, but a single example is enough to disprove a universal claim.)

I think Barbell Medicine has good logic, too. In their podcast on HIIT, they talk about how in strength training, nobody lifts the maximum possible weight every week. (And I have enough personal experience to know that maxing out every week wouldn’t work.) So it probably doesn’t make sense to max out your aerobic capacity in every week, either. I’m not sure strength training and cardio work the same way in that respect, but I expect things to be the same unless I have reason to believe they’re different.

The Barbell Medicine article on HIIT⁸ has some nice sample workouts that line up with Magness’s recommendations:

Workout #1

4 to 6 rounds of: 30 seconds on at 600-800 m running pace (or a speed sustainable in the range of ~90-150 seconds), 4 min off / easy effort

Workout #2

8 to 10 rounds of: 1-minute on at 1 mile-5 km running pace (or a speed sustainable in the range of 6-25 minutes), 1 minute off

Workout #3

3 to 5 rounds of: 5 minutes on at zone 4 heart rate (85-95% max), 3 min rest

I want to investigate this further, but here’s what I tentatively believe:

1. VO2max predicts longevity, but athletic performance matters more than VO2max alone.
2. I should exercise at a variety of intensities to get a well-rounded fitness capacity, but HIIT isn’t particularly better for improving fitness than LIT, and 4-minute intervals aren’t particularly better than other interval schemes.
3. Intervals should not be done at the maximum sustainable intensity; they should be done at an intensity that’s challenging but doesn’t leave you wiped out. As Magness wrote, “The goal isn’t to create fatigue, that’s easy to do. The goal is to slightly embarrass your body in the right direction.”

Coming back to my own training: I have loathed every version of HIIT I’ve tried so far. But that’s because I listened to the people saying that HIIT should be “all-out.” Next time I’m going to do HIIT at a more comfortable pace.

Notes