How to Increase Bone Density With Exercise: The Load Threshold Most Programs Never Reach

Most exercise prescribed for bone health is too light to change bone.

That is the uncomfortable summary of the last two decades of trials. Bone responds to large forces applied quickly. Gentle, sustained, low-force activity — the kind most often recommended to people worried about their bones — produces small or absent changes in bone mineral density (BMD).

Here is how the main options compare in postmenopausal women, the population with the most trial data:

Exercise type	Effect on spine BMD	Effect on hip BMD	Evidence strength
Walking alone	No significant effect¹	Possible small gain after 6+ months, not significant overall¹	Meta-analysis, 10 trials
Swimming or cycling	No weight-bearing stimulus; endurance athletes in these sports often have lower BMD than peers²	Same²	Narrative review
High-intensity resistance + impact	+2.9% vs −1.2% in controls over 8 months³	+0.3% vs −1.9% in controls³	Single RCT, 101 women
Resistance training (pooled)	Significant, high heterogeneity⁴	Significant; ≥70% 1RM subgroup drives hip effect⁴	Meta-analysis, 17 RCTs
Aerobic + resistance combined	Ranked highest of eight modalities⁵	Ranked highest of eight modalities⁵	Network meta-analysis, 49 studies
High-impact only (e.g. jumping alone)	Ineffective⁶	Ineffective⁶	Meta-analysis

Two things stand out. Load magnitude matters more than time spent. And no single modality is a clean winner — the strongest results come from programs that combine heavy resistance with impact.

Why bone ignores most exercise

Think of bone the way you’d think of a callus. Skin thickens where it gets rubbed hard, not where it gets touched gently for a long time. The signal is intensity, not duration.

The mechanical principle behind bone adaptation is that osteogenic loading requires high-magnitude strains applied at high rates.³ Both halves matter. A force that is large but applied slowly, and a force that is fast but small, each produce weaker signals than a force that is large and fast.

This is why duration is such a poor proxy. Ten thousand steps apply a force roughly equal to your bodyweight, ten thousand times, at a rate your skeleton adapted to before you were born. A heavy deadlift applies several times that force in under two seconds.

The American College of Sports Medicine’s position stand on physical activity and bone health, led by Wendy M. Kohrt, PhD, was built on the same premise: bone-loading exercise is classified by the mechanical forces an activity generates, not by its energy cost.⁷

Does walking increase bone density?

Walking is the most commonly recommended exercise for bone health. It is also the least supported by the data.

Ma and colleagues pooled ten trials of walking interventions in perimenopausal and postmenopausal women. At the lumbar spine, walking produced no significant effect on BMD regardless of how long the intervention ran (weighted mean difference 0.01 g/cm², 95% CI −0.00 to 0.02, P = 0.05). Femoral neck BMD improved in trials running six months to two years, but the effect across all trials did not reach significance (P = 0.07). Effects at the radius and whole body were not significant.¹

This is not an argument against walking. Walking has cardiovascular, metabolic, and mood benefits that are well established, and we’ve covered what step counts and heart-rate zones actually do for weight management. It is an argument against expecting walking to protect your skeleton.

What about swimming and cycling?

No — both remove the variable bone responds to, which is bodyweight and load travelling through the skeleton. Scofield and Hecht’s review of bone health in endurance athletes found that adults in non-weight-bearing sports such as cycling and swimming often have lower BMD than athletes in ball and power sports — and sometimes lower than their inactive peers.²

An athlete can be extremely fit and have a fragile skeleton. Cardiovascular fitness and bone strength are separate adaptations that require separate stimuli — which is why SensAI treats a week of cycling and a week of heavy lifting as different training loads rather than interchangeable minutes of exercise.

What the intensity data shows

The clearest test of the load-magnitude idea came from Kistler-Fischbacher, Weeks, and Beck, who meta-analysed 53 trials in postmenopausal women and grouped interventions by intensity rather than by modality.

At the lumbar spine, high-intensity exercise produced a mean difference of 0.031 g/cm² (95% CI 0.012 to 0.049, p = 0.002) — roughly two to three times the effect of moderate intensity (0.012 g/cm²) or low intensity (0.010 g/cm²).⁸

The femoral neck told a different story. Low and moderate intensity were equally effective there (both 0.011 g/cm²), and no effect of high-intensity exercise was observed. The authors were candid about why: of 63 interventions analysed, only four were high intensity. The null result at the hip may reflect a lack of statistical power rather than a lack of effect.⁸

That caveat is worth holding onto. The high-intensity evidence base is strong in direction and thin in volume.

Can people with osteoporosis lift heavy? The LIFTMOR trials

The most-cited demonstration of heavy loading for bone is the LIFTMOR trial, run by Watson and colleagues at Griffith University with Belinda R. Beck, PhD — Professor at Griffith’s Menzies Health Institute Queensland and Director of The Bone Clinic in Brisbane — as senior author.

The design was deliberately provocative. High-intensity resistance and impact training (HiRIT) had traditionally been withheld from people with osteoporosis on the assumption that it would fracture them. LIFTMOR tested it anyway.

LIFTMOR protocol	Detail
Participants	101 postmenopausal women with low bone mass (T-score < −1.0, meaning at least one standard deviation below a young-adult reference), aged 65 ± 5 years
Duration	8 months
Frequency	Twice weekly
Session length	30 minutes, supervised
Prescription	5 sets of 5 repetitions at >85% of 1 repetition maximum
Control	Home-based, low-intensity exercise

Results favoured HiRIT at every measured site. Lumbar spine BMD rose 2.9 ± 2.8% in the training group while falling 1.2 ± 2.8% in controls (p < 0.001). Femoral neck BMD rose 0.3 ± 2.6% versus a 1.9 ± 2.6% decline in controls (p = 0.004). Femoral neck cortical thickness, height, and every functional performance measure also favoured HiRIT.³

Adverse events: one. A minor lower back spasm that cost two missed sessions out of 70. Compliance was 92 ± 11%.³

The companion trial in men, LIFTMOR-M, produced a similar pattern across 93 men averaging 67 years. HiRIT improved trochanteric BMD by 2.8 ± 0.8% against −0.1 ± 0.9% in controls (p = .024) and lumbar spine BMD by 4.1 ± 0.7% against 0.9 ± 0.8% (p = .003), with five minor adverse events.⁹

Read those numbers carefully. The training groups gained a few percent. The control groups lost bone. Much of the benefit is loss prevented, not loss reversed — which is still the outcome that matters when the alternative is a hip fracture.

One structural point: LIFTMOR was supervised, every session, by exercise professionals. The safety record belongs to the supervision as much as to the protocol.

How much load, how often

Zhao and colleagues pooled 17 RCTs covering 690 postmenopausal women to isolate the training variables that matter. Resistance training significantly improved BMD at the lumbar spine (standardised mean difference, SMD, of 0.88, 95% CI 0.21 to 1.56 — a large effect), femoral neck (SMD 0.89, 95% CI 0.40 to 1.39), and total hip (SMD 0.30, 95% CI 0.10 to 0.50), but not the trochanter.⁴

Their subgroup findings pointed in a consistent direction:

Intensity: high-intensity training at ≥70% of 1RM drove the significant effects at the total hip and femoral neck.
Frequency: three sessions per week improved BMD at every site measured.
Duration: interventions of 48 weeks or longer were needed for femoral neck and total hip effects.

The authors also flagged heterogeneity of I² = 91% at the spine and 87% at the femoral neck — the trials disagreed with each other far more than chance explains — and said plainly that it limits how far the pooled estimates generalise.⁴ Treat these as direction, not dosage.

The 48-week finding deserves emphasis. Bone remodelling is slow. A 12-week program that transforms your strength will do very little to your DXA scan (the X-ray that measures bone density), and a scan repeated too soon will mislead you about whether the program is working. This is the opposite of the feedback loop most training apps are built around, and it’s one reason SensAI frames progression around load and rep history rather than short-term outcome metrics — the training variable is what you can actually observe week to week.

Resistance alone or resistance plus impact?

Here the evidence gets less tidy.

Martyn-St James and Carroll found that mixed loading programs worked and single-mode ones did not. Protocols combining jogging with walking and stair climbing, and protocols mixing impact with high-magnitude resistance exercise, improved spine and femoral neck BMD. High-impact-only and odd-impact-only protocols were ineffective at every site.⁶

A 2025 network meta-analysis by Xiaoya and colleagues covering 49 studies and 3,360 participants across eight interventions ranked aerobic plus resistance training highest for both lumbar spine and femoral neck BMD, ahead of resistance training alone and whole-body vibration.⁵

The 2011 Cochrane review by Howe and colleagues, pooling 43 RCTs and 4,320 participants, reached a compatible conclusion by a different route: the most effective intervention for femoral neck BMD was non-weight-bearing high-force exercise such as progressive resistance strength training for the lower limbs (MD 1.03, 95% CI 0.24 to 1.82), while the most effective for spine BMD was combination programs (MD 3.22, 95% CI 1.80 to 4.64).¹⁰

Different methods, same shape of answer. Combine loading types. Do not rely on one.

The finding nobody advertises

The Cochrane review found no effect of exercise on the number of fractures (odds ratio 0.61, 95% CI 0.23 to 1.64).¹⁰

That confidence interval crosses 1.0, which means the trials could not distinguish a large benefit from a modest harm. Fractures are rare events, and the trials were not designed or powered to detect changes in them. Absence of evidence is not evidence of absence — but it is also not the fracture-prevention claim that often gets attached to this literature.

BMD is a surrogate. It correlates with fracture risk; it is not the same thing.

Which is why the falls evidence carries as much practical weight as the density evidence. Cathie Sherrington, PhD, Professor at the University of Sydney’s Institute for Musculoskeletal Health, led the Cochrane review of exercise for falls prevention: exercise reduces the rate of falls by 23% (rate ratio 0.77, 95% CI 0.71 to 0.83; 12,981 participants across 59 studies, high-certainty evidence). Balance and functional exercises reduce falls by 24%. Programs combining multiple exercise types — commonly balance and functional work plus resistance training — probably reduce falls by 34%.¹¹

Notably, the review was uncertain about the effect of programs consisting primarily of resistance training.¹¹

Put the two literatures together and the practical target is clear: heavy loading to defend bone density, plus balance and functional work to reduce the number of times you land on it. Most people fracture because they fall, not because a bone spontaneously fails.

Building an exercise program for osteopenia and osteoporosis

The gap between “lifting weights” and “lifting weights heavy enough to change bone” is where most well-intentioned programs stall.

Exercise and Sports Science Australia’s position statement — authored by Beck, Robin M. Daly, Maria A. Fiatarone Singh, and Dennis R. Taaffe — sets out the practical guardrails:¹²

Bone responds positively to impact activities and high-intensity progressive resistance training.
Optimising muscle strength, balance, and mobility reduces fall risk, which matters most for people with limited functional capacity or very high fracture risk.
Loaded spine flexion is not recommended.
Impact activities may need modification in the presence of osteoarthritis or frailty.
All programs should be accompanied by sufficient calcium and vitamin D.
Prescription must account for existing bone status, comorbidities, and fall risk.

That last point is the one that resists automation. The same protocol that produced one minor adverse event in LIFTMOR’s screened, supervised cohort is not automatically appropriate for someone with vertebral fractures, uncontrolled hypertension, or advanced osteoarthritis. If you have diagnosed osteopenia or osteoporosis, this is a conversation with your physician and ideally a supervised start with an exercise professional — not a program you download.

For people cleared to train, the practical build is unglamorous:

Learn the five loadable patterns before adding weight. Squat (back squat, goblet squat), hinge (deadlift, Romanian deadlift), push (overhead press), pull (row, lat pulldown), carry (farmer’s walk). Technique under light load first — and if loaded training is new to you, our beginner’s guide to strength training for women covers the on-ramp in detail.
Progress until the load is genuinely heavy. The trial evidence clusters at ≥70% 1RM, with LIFTMOR at >85%. Sets of 5 to 8 hard repetitions are the working range, not sets of 15 comfortable ones.
Add impact deliberately. Jumping, hopping, or drop landings — but as part of a mixed program, since impact alone did not work.⁶
Train three times a week and measure in years. Zhao’s data favoured three weekly sessions and durations beyond 48 weeks.⁴
Keep balance and functional work in the program. It carries the strongest fracture-relevant evidence.¹¹

Progressive overload sounds simple and is where most self-directed programs quietly fail — the load stops climbing and the stimulus stops mattering. We’ve written separately about how to structure progressive overload so the load actually keeps moving. SensAI’s conversational coach tracks working loads across sessions and can flag when a lift has plateaued, which is the difference between a program that stays heavy and one that only feels hard.

Where this applies beyond postmenopausal women

Almost all of the trial evidence above comes from postmenopausal women, and to a lesser extent older men via LIFTMOR-M. That is a real limitation.

The mechanical principle — high-magnitude, high-rate loading — is not sex-specific or age-specific, and peak bone mass is built in adolescence and early adulthood, which means the highest-leverage window closes before most people start worrying about it. But the effect sizes measured in 65-year-old women with low bone mass should not be assumed to transfer to a 30-year-old with normal density.

If you’re training through the menopause transition, when bone loss accelerates, we’ve covered how to adjust training through perimenopause in more detail. For readers newer to loaded training, our guide to what actually changes about training after 40 covers how to adapt the programming.

SensAI builds programs from your training history, connected wearable data, and stated constraints rather than a fixed template, which matters here because bone-focused training is defined by what it excludes as much as what it includes — loaded spinal flexion, impact work that a given joint cannot tolerate, progressions that outrun technique. Those are constraints you should be able to state in plain language and have respected, and that’s what the conversational coach is for.

What we still don’t know

Being honest about the edges of this evidence:

The high-intensity trial base is small. Four high-intensity interventions out of 63 in the largest intensity meta-analysis.⁸
Fracture outcomes remain unproven. The Cochrane confidence interval spans benefit and harm.¹⁰
Heterogeneity is high. I² above 85% in the resistance-training pooled estimates.⁴
Supervision is a confounder. LIFTMOR’s safety record came with expert supervision at every session.³
Hip responds less than spine. Across multiple analyses, femoral neck gains are smaller and less consistent than lumbar spine gains.³¹⁰⁸

None of that undermines the central finding, which is unusually consistent across methods: bone responds to heavy loading and largely ignores light loading. It does mean the honest claim is “heavy resistance training preserves and modestly improves bone density,” not “exercise prevents fractures.”

The practical implication is the same either way. If bone health is why you’re exercising, the load has to get heavy enough to matter, the program has to run for years rather than weeks, and it should include balance work — because the fracture you avoid is most likely the fall you never took.

References

Ma D, Wu L, He Z. “Effects of walking on the preservation of bone mineral density in perimenopausal and postmenopausal women: a systematic review and meta-analysis.” Menopause, 2013;20(11):1216-1226. https://pubmed.ncbi.nlm.nih.gov/24149921/ ↩ ↩² ↩³
Scofield KL, Hecht S. “Bone health in endurance athletes: runners, cyclists, and swimmers.” Current Sports Medicine Reports, 2012;11(6):328-334. https://pubmed.ncbi.nlm.nih.gov/23147022/ ↩ ↩² ↩³
Watson SL, Weeks BK, Weis LJ, Harding AT, Horan SA, Beck BR. “High-Intensity Resistance and Impact Training Improves Bone Mineral Density and Physical Function in Postmenopausal Women With Osteopenia and Osteoporosis: The LIFTMOR Randomized Controlled Trial.” Journal of Bone and Mineral Research, 2018;33(2):211-220. https://pubmed.ncbi.nlm.nih.gov/28975661/ ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷
Zhao F, Su W, Sun Y, Wang J, Lu B, Yun H. “Optimal resistance training parameters for improving bone mineral density in postmenopausal women: a systematic review and meta-analysis.” Journal of Orthopaedic Surgery and Research, 2025;20(1):523. https://pubmed.ncbi.nlm.nih.gov/40420105/ ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶
Xiaoya L, Junpeng Z, Li X, Haoyang Z, Xueying F, Yu W. “Effect of different types of exercise on bone mineral density in postmenopausal women: a systematic review and network meta-analysis.” Scientific Reports, 2025;15(1):11740. https://pubmed.ncbi.nlm.nih.gov/40188285/ ↩ ↩² ↩³
Martyn-St James M, Carroll S. “A meta-analysis of impact exercise on postmenopausal bone loss: the case for mixed loading exercise programmes.” British Journal of Sports Medicine, 2009;43(12):898-908. https://pubmed.ncbi.nlm.nih.gov/18981037/ ↩ ↩² ↩³ ↩⁴
Kohrt WM, Bloomfield SA, Little KD, Nelson ME, Yingling VR. “American College of Sports Medicine Position Stand: physical activity and bone health.” Medicine & Science in Sports & Exercise, 2004;36(11):1985-1996. https://pubmed.ncbi.nlm.nih.gov/15514517/ ↩
Kistler-Fischbacher M, Weeks BK, Beck BR. “The effect of exercise intensity on bone in postmenopausal women (part 2): A meta-analysis.” Bone, 2021;143:115697. https://pubmed.ncbi.nlm.nih.gov/33357834/ ↩ ↩² ↩³ ↩⁴
Harding AT, Weeks BK, Lambert C, Watson SL, Weis LJ, Beck BR. “A Comparison of Bone-Targeted Exercise Strategies to Reduce Fracture Risk in Middle-Aged and Older Men with Osteopenia and Osteoporosis: LIFTMOR-M Semi-Randomized Controlled Trial.” Journal of Bone and Mineral Research, 2020;35(8):1404-1414. https://pubmed.ncbi.nlm.nih.gov/32176813/ ↩
Howe TE, Shea B, Dawson LJ, Downie F, Murray A, Ross C, Harbour RT, Caldwell LM, Creed G. “Exercise for preventing and treating osteoporosis in postmenopausal women.” Cochrane Database of Systematic Reviews, 2011;(7):CD000333. https://pubmed.ncbi.nlm.nih.gov/21735380/ ↩ ↩² ↩³ ↩⁴
Sherrington C, Fairhall N, Wallbank G, Tiedemann A, Michaleff ZA, Howard K, Clemson L, Hopewell S, Lamb S. “Exercise for preventing falls in older people living in the community: an abridged Cochrane systematic review.” British Journal of Sports Medicine, 2020;54(15):885-891. https://pubmed.ncbi.nlm.nih.gov/31792067/ ↩ ↩² ↩³
Beck BR, Daly RM, Singh MA, Taaffe DR. “Exercise and Sports Science Australia (ESSA) position statement on exercise prescription for the prevention and management of osteoporosis.” Journal of Science and Medicine in Sport, 2017;20(5):438-445. https://pubmed.ncbi.nlm.nih.gov/27840033/ ↩