The short version: in 2024 the Endocrine Society — the body that gave the world the 75 nmol/L vitamin D target in 2011 — formally abandoned both that target and routine testing. Its guideline states that it "no longer endorses specific 25(OH)D levels to define vitamin D sufficiency, insufficiency, and deficiency."

That is a remarkable thing for a professional body to say about a test performed millions of times a year, and there are two good reasons for it.

The assays disagree. When Australian researchers sent the same 50 serum samples to three laboratories and a certified reference lab, the measured deficiency prevalence came out at 24% (reference), 16%, 12% and 41%. Two labs read about 12 nmol/L high; one read 10.6 nmol/L low. Your "deficiency" may be your laboratory's choice of instrument.

And the big trials found nothing. Australia's own D-Health trial gave 60,000 IU monthly to 21,315 people over 60 for five years and found an all-cause mortality hazard ratio of 1.04 (95% CI 0.93–1.18) — if anything, slightly the wrong way.

Also worth knowing before you buy anything: the Australian upper limit is 80 µg/day, which is 3,200 IU — not the 4,000 IU you may have read. That is a US figure.

The Medicare restriction, and why it exists

Australia once had the world's most enthusiastic vitamin D testing habit. Testing went from 0.4 tests per 1,000 population in 2000 to 36.5 per 1,000 in 2011, with the Medicare bill rising from about $1.1 million to $95.6 million over the same decade, peaking around $151 million in 2012–13.

On 1 November 2014, items 66608 and 66609 were replaced by items 66833–66837, with eligibility criteria attached.

MBS item 66833 — the actual criteria

Fee $30.05; benefit 75% = $22.55, 85% = $25.55. "25-hydroxyvitamin D, quantification in serum, for the investigation of a patient who:"

• (a) has signs or symptoms of osteoporosis or osteomalacia; or
• (b) has increased alkaline phosphatase and otherwise normal liver function tests; or
• (c) has hyperparathyroidism, hypo- or hypercalcaemia, or hypophosphataemia; or
• (d) is suffering from malabsorption (for example, because the patient has cystic fibrosis, short bowel syndrome, inflammatory bowel disease or untreated coeliac disease, or has had bariatric surgery); or
• (e) has deeply pigmented skin, or chronic and severe lack of sun exposure for cultural, medical, occupational or residential reasons; or
• (f) is taking medication known to decrease 25OH-D levels (for example, anticonvulsants); or
• (g) has chronic renal failure or is a renal transplant recipient; or
• (h) is less than 16 years of age and has signs or symptoms of rickets; or
• (i) is an infant whose mother has established vitamin D deficiency; or
• (j) is a exclusively breastfed baby and has at least one other risk factor mentioned in a paragraph in this item; or
• (k) has a sibling who is less than 16 years of age and has vitamin D deficiency

(The typo in (j) is in the source.)

Note what is not on that list: feeling tired, working indoors, being over 50, or general wellness. If your GP orders a vitamin D test without one of these, you may be paying for it yourself.

What happened next

**Gordon L, Waterhouse M, Reid IR, Neale RE. "The vitamin D testing rate is again rising, despite new MBS testing criteria." Med J Aust 2020;213(4):155.**

• The testing rate was 47% lower during 2014–16 than during 2013–14
• Then from 2015 to 2019 it rose 34%, from 119 to 159 tests per 1,000 population
• Women: 164 → 214 per 1,000 (+30%). Men: 74 → 105 (+40%)
• Medicare cost: $73.7 million in 2015 → $104.7 million in 2019 (+42%)

Be careful with both percentages. The 47% compares two multi-year periods rather than consecutive years, and the 34% rise starts from an already-suppressed post-restriction base. The honest summary: the restriction worked, and then it stopped working. Australia is now testing more than it was immediately after the rules were tightened, under rules that have not changed.

An earlier analysis (MJA 2016;204(1)) recorded Medicare benefits falling from about $143.1 million in 2013–14 to $101.2 million in 2014–15, roughly 29%, and projected ongoing savings of about $64 million a year.

The assay problem

This is the part that should give anyone pause before acting on a single number.

**"Analytical bias in the measurement of serum 25-hydroxyvitamin D concentrations impairs assessment of vitamin D status in clinical and research settings." PLoS ONE 2015** (Australian authors, Telethon Kids / ANU / UTAS).

Design: 50 participants' serum, measured by three laboratories and compared against a certified reference laboratory.

(The deficiency threshold is almost certainly <50 nmol/L but we could not confirm it from the source we read. We also could not capture the funding statement, or the volume and DOI — the PubMed ID appears to be 26266807.)

Same blood. Same fifty people. Deficiency prevalence between 12% and 41% depending on which lab opened the tube.

The authors' recommendation: "all laboratories should use an accuracy-based quality assurance system and, ideally, comply with international standardisation efforts."

Note also what this corrects. A version of this finding circulates as "four labs found 12% versus 41%." It was three labs plus a reference lab, and the reference lab's own answer — 24% — sat in the middle. That detail matters, because it means one lab roughly halved the true prevalence and another nearly doubled it.

We did not research DEQAS (the international external quality assessment scheme) or the NIH Vitamin D Standardization Program, and we are not going to summarise them from memory. If you want to know whether your pathology provider participates, ask them.

What the big trials found

D-Health — the Australian one

**Neale RE, et al. "The D-Health Trial: a randomised controlled trial of the effect of vitamin D on mortality." Lancet Diabetes Endocrinol 2022;10(2):120–128.**

n=21,315 (10,662 vitamin D, 10,653 placebo), aged 60 and over, double-blind, 60,000 IU monthly for 5 years, median follow-up 5.7 years. 1,100 deaths.

Deaths: 5.1% placebo, 5.3% vitamin D. Exploratory analyses suggested a possible increase in cancer mortality later in follow-up — the authors flagged this as exploratory and it should be treated that way, not as a finding.

This is the largest Australian vitamin D trial, in the population most often told to supplement, and it found nothing. (Funding was not captured in our source; we believe NHMRC. We also did not retrieve the fracture and falls substudies.)

VITAL — the American one

**Manson JE, et al. NEJM 2019;380(1):33–44. n=25,871**, 2,000 IU/day vitamin D3, 2×2 factorial with omega-3, median follow-up 5.3 years, men 50+ and women 55+. Funded by multiple NIH institutes.

A post-hoc analysis excluding the first two years showed lower cancer mortality, without multiplicity adjustment. That is a hypothesis generated after the fact in a trial that missed its primary endpoints.

The bone substudy was null. LeBoff MS, et al. NEJM 2022;387:299–309 — 1,991 incident fractures among 1,551 participants:

• Total fractures: HR 0.98 (0.89–1.08), p=0.70
• Non-vertebral: HR 0.97 (0.87–1.07), p=0.50
• Hip: HR 1.01 (0.70–1.47), p=0.96
• Results consistent across baseline 25(OH)D, age, sex, race and BMI
• No effect on falls, bone mineral density or bone structure

That last line is the important one. The reason most people take vitamin D is bones. In 25,871 people over five years, at 2,000 IU a day, it did nothing to fractures — including in people who started with low levels.

The autoimmune substudy was the one positive result. Hahn J, et al. BMJ 2022: vitamin D alone HR 0.78 (95% CI 0.61–0.99), with 123 cases versus 155. Omega-3 alone showed a 15% reduction, p=0.19, not significant.

Read that confidence interval: the upper bound is 0.99. It is a secondary analysis with a result that barely clears significance. We could not confirm whether it was pre-specified as an ancillary study, or verify the case counts' arm assignment. It is the most interesting positive signal in the modern vitamin D literature and it is nowhere near strong enough to act on.

The fracture meta-analysis

**Bolland MJ, Grey A, Avenell A. Lancet Diabetes Endocrinol, published online 4 October 2018. 81 RCTs, 53,537 participants.** Funded by the Health Research Council of New Zealand.

Supplementation did not reduce total fracture, hip fracture or falls, and was not associated with clinically relevant bone mineral density improvement at any site. Trial sequential analysis suggested vitamin D might increase hip fracture risk — suggested, not demonstrated.

The authors' conclusion: "little justification for the use of vitamin D supplements to maintain or improve musculoskeletal health."

We did not retrieve the individual effect sizes and confidence intervals and are not printing them.

The high-dose harm signal — also Australian

**Sanders KM, et al. JAMA 2010. 2,256 women aged 70+ at high fracture risk, placebo-controlled, annual oral 500,000 IU cholecalciferol**, 3–5 years. Funded by the NHMRC and the Australian Government Department of Health and Ageing.

• Serum 25(OH)D rose by 41% on average
• Falls: 83 vs 73 per 100 person-years — a 15% relative increase
• Fractures: 4.9 vs 3.9 per 100 person-years (171 fractures in 155 women vs 135 in 125) — a 26% relative increase

More vitamin D produced more fractures. (We did not capture the rate ratios, confidence intervals or p-values and are not stating them. The citation we have is JAMA 2010;303(18):1815–1822, from memory rather than from the page.)

This was an annual megadose, a different regimen from D-Health's monthly dosing — but it is the clearest demonstration available that the dose-response curve for vitamin D is not a straight line upward. It also underlies the 2024 Endocrine Society preference for daily lower-dose over intermittent high-dose supplementation.

The respiratory infection story, and how a subgroup evaporated

This is worth following closely because it is one of the cleanest examples in medicine of a compelling subgroup finding that did not survive.

2017: a striking subgroup

**Martineau AR, et al. "Vitamin D supplementation to prevent acute respiratory tract infections: systematic review and meta-analysis of individual participant data." BMJ 2017.** 25 RCTs, 11,321 participants, individual participant data for 10,933 (96.6%).

Serious adverse events: aOR 0.98 (0.80–1.20), p=0.83.

A 70% reduction in respiratory infection in the severely deficient. That number went around the world, and reasonably so — it is biologically plausible, the analysis was individual-participant, and the dosing-frequency pattern made mechanistic sense.

2025: it didn't hold

**Jolliffe DA, et al. Lancet Diabetes Endocrinol 2025;13:307–320. A meta-analysis of stratified aggregate data combining the 43 trials from the 2021 analysis (48,488 participants) with 6 newly completed trials (19,337 participants) published 2020–2024. Around 61,589 participants** in total.

• Updated pooled OR 0.94 (95% CI 0.88–1.00), p=0.057
• The 2021 estimate had been OR 0.92 (0.86–0.99)
• Prespecified subgroup analyses found "no evidence of effect modification by age, baseline vitamin D status, dosing frequency, or dose size"
• Serious adverse events: OR 0.96 (0.90–1.04)
• A funnel plot suggested small-study effects

So the overall effect shrank until it no longer reached significance, and every subgroup pattern — including the 0.30 in severe deficiency and the daily-versus-bolus distinction — disappeared.

Two honest caveats. The designs differ: 2017 was individual-participant data, 2025 was stratified aggregate data, and the latter is less sensitive to effect modification. So this is not the same subgroup test run twice on the same data and failing. And the figure of "about 0.98" for the 2025 result, which circulates (including in our own earlier notes), is wrong — it was 0.94 with a CI touching 1.00.

What survives: vitamin D probably does very little for respiratory infections in most people, and the spectacular subgroup effect has not replicated. Six new trials with 19,337 participants moved the needle the wrong way, and a funnel plot suggests the earlier literature was skewed by small studies.

So who should actually take it?

The guideline that matters most now is the 2024 Endocrine Society clinical practice guideline (Demay MB, et al. J Clin Endocrinol Metab 2024;109(8):1907–1947). Its recommendations, with the Society's own strength and certainty ratings:

Routine 25(OH)D testing is recommended against in every one of those groups, rated 2 / ⊕◯◯◯.

Every recommendation is conditional (strength "2"), and none exceeds moderate certainty. Note the structure the Society arrived at: supplement some groups without testing them, and don't test anyone routinely. That is close to the opposite of how vitamin D is handled in practice, where people get tested and then supplemented to a target.

Our source for this table was a summarised PDF and the numbered list it returned does not obviously match the published guideline's structure. Check the original before quoting the wording.

And the threshold statement, verbatim: "Endocrine Society no longer endorses the target 25(OH)D level of 30 ng/mL (75 nmol/L) suggested in the previous guideline." The panel "no longer endorses specific 25(OH)D levels to define vitamin D sufficiency, insufficiency, and deficiency."

We were not able to retrieve the current Australian position — the 2012 MJA position statement, which uses a 50 nmol/L threshold, or the current Healthy Bones Australia / ANZBMS statement. The Australian position has not changed to match the 2024 guideline as far as we know, and 50 nmol/L remains the working threshold here. Do not assume your GP is applying the Endocrine Society's new position.

The 600 IU argument, for completeness

**Veugelers PJ, Ekwaru JP. "A Statistical Error in the Estimation of the Recommended Dietary Allowance for Vitamin D." Nutrients 2014;6(10):4472–4475.**

The US Institute of Medicine claimed 600 IU/day brings 97.5% of individuals to ≥50 nmol/L. The authors argue the lower prediction limit used applies to study averages, not individuals, and recalculate: 600 IU/day gives a 2.5th percentile of only 26.8 nmol/L, and reaching 50 nmol/L in 97.5% of individuals would require about 8,895 IU/day.

This paper is cited constantly by the high-dose end of the supplement industry, so read it carefully. The statistical criticism may well be right. The 8,895 IU figure is an extrapolation from a reanalysis, not a trial finding, it sits far above the Australian upper limit, and the trials that have actually tested high doses — Sanders 2010, D-Health — found harm or nothing. A published response to the paper exists which we did not read.

Sun, and the Australian trade-off

Australia has an unusual problem: the same ultraviolet radiation that makes vitamin D causes the skin cancers we lead the world in.

Cancer Council Australia's position, as published:

• "Sensible sun protection does not put people at risk of vitamin D deficiency."
• Northern Australia: protect whenever the UV Index is 3 or above. Unprotected time is possible early morning and late afternoon when it is below 3.
• Southern Australia: protect in summer and early autumn at UV 3+. In late autumn and winter, when UV stays below 3, protection may not be needed — spend time outdoors over the middle of the day with some skin uncovered.
• In summer, "a few minutes outdoors on most days" is enough.
• At-risk groups — dark skin, routinely covered skin, elderly or institutionalised people — "should consult doctors about supplementation rather than increasing sun exposure."

The NHMRC's nutrient reference material describes roughly 15–30 minutes, two to three times weekly as sun exposure equivalent to dietary intake, varying by skin type, latitude and season.

We could not retrieve a verified minutes-by-latitude-and-season table. Healthy Bones Australia publishes a "sunshine map" but the page returned a 404 when we checked. If you want numbers for your city and season, that is the source to look for — we are not going to invent them, and the correct answer genuinely does depend on your skin type, the season, the time of day and how much skin is uncovered.

Dietary sources and fortification: we could not verify what is fortified in Australia — the FSANZ page we tried returned a 404. Healthdirect states only that "most people cannot get enough vitamin D from food alone," which is true and not very useful. Our recollection is that there is a mandatory requirement for table spreads and voluntary fortification of some dairy, but we have not verified it and you should not rely on that sentence.

On prevalence: the 2016 MJA piece cited estimates that 31–58% of Australians are deficient but only about 4% moderately to severely so. Treat that as a quoted estimate, not a measurement — and recall from the assay section that the number depends heavily on which laboratory produced it.

The upper limit, and it is not 4,000 IU

From the NHMRC Nutrient Reference Values (eatforhealth.gov.au):

The Australian and New Zealand upper limit is 3,200 IU/day. The US figure is 4,000 IU. If you have read 4,000 IU in an Australian context, it came from an American source. (The conversion is 40 IU per microgram: 80 × 40 = 3,200.)

Two things follow. A 5,000 IU capsule — widely sold — exceeds the Australian upper level by more than 50%. And the common "specialist" weekly products at 7,000 IU average 1,000 IU a day, which is comfortably within it.

We did not research hypercalcaemia or megadose toxicity case reports. The high-dose trial signals in the Sanders and D-Health sections above are the data we have.

What to do

Don't get tested unless you meet an MBS criterion. The list is above. If you don't, you are paying for a measurement that two professional bodies now advise against doing routinely, and that three laboratories would answer differently.

If you are over 75, pregnant, or have high-risk prediabetes, supplementation is reasonable without testing. That is the 2024 guideline's position, at conditional strength. Daily, modest dose.

If you are under 50 and otherwise well, the guideline recommends against supplementing beyond ordinary dietary intake. The trials in your situation found nothing.

Don't take it for your bones on the strength of the marketing. In 25,871 people over five years at 2,000 IU/day, it did nothing to fractures, falls, bone density or bone structure — including in those who started low. An 81-trial meta-analysis in 53,537 people reached the same conclusion.

Don't take megadoses. An annual 500,000 IU in Australian women produced 15% more falls and 26% more fractures. Daily and modest beats intermittent and large, and that is now in the guideline.

Don't exceed 3,200 IU a day without a doctor's reason. That is the Australian upper level, not 4,000.

If you have genuinely low sun exposure — covered clothing, night shift, institutional care, dark skin at high latitude — talk to a GP about supplementing rather than about sunbathing. That is Cancer Council's advice and it is the group where supplementation has the clearest case.

And if you are already on it and feel fine, there is no harm in a modest dose. 1,000 IU a day costs cents, sits well inside the upper limit, and the trials that found nothing also found nothing bad. This article is an argument against testing and against megadoses, not an argument that you must stop.

Frequently asked

Should I get my vitamin D tested? Only if you meet one of the Medicare criteria — osteoporosis or osteomalacia symptoms, abnormal calcium or alkaline phosphatase, malabsorption, deeply pigmented skin or severe lack of sun exposure, certain medications, kidney failure, or specific paediatric situations. The 2024 Endocrine Society guideline recommends against routine testing in every group it considered, and when the same 50 samples went to three Australian labs plus a reference lab, measured deficiency came out at 12%, 16%, 24% and 41%.

What vitamin D level is normal? The honest answer is that the field no longer agrees. The Endocrine Society explicitly withdrew its 75 nmol/L target in 2024 and declined to endorse any threshold. The working Australian figure remains 50 nmol/L. Given the assay variation above, a result near any threshold should be read as approximate.

Does vitamin D prevent fractures? The best evidence says no. VITAL randomised 25,871 people to 2,000 IU/day for five years and found no effect on total fractures (HR 0.98), hip fractures (HR 1.01), falls, bone mineral density or bone structure — including in people with low baseline levels. A meta-analysis of 81 trials in 53,537 people concluded there is "little justification" for using it to improve musculoskeletal health.

Does vitamin D prevent colds and flu? Probably very little. A 2017 individual-participant meta-analysis found an overall odds ratio of 0.88 and a striking 0.30 in people with severe deficiency. A 2025 update adding six new trials and 19,337 participants found a pooled OR of 0.94 with a confidence interval touching 1.00 (p=0.057), and no effect modification by baseline vitamin D status, dose or dosing frequency. The subgroup effect did not replicate.

How much sun do I need? It depends on your skin type, latitude, season, time of day and how much skin is uncovered, which is why no single number is honest. Cancer Council Australia's position is that sensible sun protection does not put people at risk of deficiency, that in summer "a few minutes outdoors on most days" suffices, and that in southern Australia in late autumn and winter — when the UV index stays below 3 — protection may not be needed and midday outdoor time with some skin uncovered is appropriate.

How much is too much? The Australian and New Zealand upper level of intake is 80 µg/day, which is 3,200 IU — not the 4,000 IU figure from the United States. An Australian trial of annual 500,000 IU doses in women over 70 found 15% more falls and 26% more fractures than placebo, so more is demonstrably not better.

Is 5,000 IU a day safe? It exceeds the Australian upper level by more than 50%. We have no toxicity data to offer at that specific dose, and plenty of trial data showing no benefit at a quarter of it. There is no evidence-based reason to take it unless a doctor is managing a documented deficiency and monitoring you.

Last reviewed 1 October 2026. Funding is noted on every trial where the source discloses it. MBS item 66833's criteria are quoted verbatim from MBS Online as at the date of review; the fee and criteria can change. Several figures we expected to print — the Sanders rate ratios, the Bolland effect sizes, Australian retail prices and a sun-exposure table by latitude — could not be verified from primary sources and have been left out rather than estimated. Two claims in our own earlier notes were wrong and are corrected above: the 2025 respiratory meta-analysis found OR 0.94, not 0.98, and the laboratory comparison involved three labs plus a reference lab, not four labs. This page is general information, not medical advice.