The Cleanest Air in Johannesburg Is 5× Too Dirty

For half the year, the air in my leafy northern-suburbs backyard is merely mediocre. For the other half — Johannesburg’s cold, still winter — it turns genuinely dirty, and not by a little. A typical winter month averages four to five times the level the World Health Organization calls safe; this July, month-to-date, my station has averaged more than seven times it. Blend the gentle summer back in and the annual average still lands at 3.5× the WHO guideline — but that tidy year-round figure is a truce between a clean summer and a punishing winter, and it’s the winter your lungs actually live through.

Breathe air like this and, over a decade, the epidemiology stops being abstract: measurably stiffer arteries, a higher lifetime risk of heart attack and stroke, a small but permanent tax on how long you live. Now picture a child growing up in it, lungs still under construction, breathing faster and closer to the ground than you do. And this — leafy, affluent, monitored — is the good part of the city.

I didn’t know any of this when I started. What I knew was that my weather app kept flashing air-quality alerts I’d always dismissed, and that a city of almost five million people had almost no independent air monitoring — a handful of government stations for a metro the size of greater London. I wanted to know if the alerts meant anything for me, specifically, at my house. The only way to find out was to measure it. So I did, for a year.

🔗 The experiment

I host an IQAir monitor — a sensor that reads fine particulate matter minute by minute and logs it to a database I run at home. It’s live and public; you can watch my suburb’s air in real time on IQAir. I’m one of only a few private stations doing this in Johannesburg, which turns out to be a large part of the story, and one I’ll come back to.

The station has sat in two northern suburbs this year: Blairgowrie, then Oaklands after I moved. Estate-agent country — the addresses people pay a premium for precisely because they assume the air, like everything else, comes clean.

Here is what a full year actually looked like.

The headline is the annual average: 17.4 micrograms of PM2.5 per cubic metre — about 3.5× the WHO’s guideline of 5 µg/m³.^[1] But as the chart shows, that average blends two very different seasons. Through summer my station sat at a fairly innocent 11–13. Then winter shut in and it roughly doubled — averaging around 22 µg/m³ from May to September, peaking at 27.6 in June, and climbing higher still into the current winter (off the right edge of this chart, and visible in the calendar further down). The annual number is real, but it’s the winter half of the year that does the damage.

PM2.5, if it’s new to you, is the fraction of pollution small enough to slip past your body’s defences, lodge deep in the lungs, and cross into the bloodstream — twenty to thirty times thinner than a human hair. It does more damage to human health than any other common pollutant, which is why it’s the number worth watching.^[2]

🔗 How dirty is that, actually?

A number like 17.4 means nothing without a yardstick, so here’s where it sits on the planet. Cities like London and New York run at roughly 7–8 µg/m³ a year; Sydney sits closer to 5. Johannesburg’s ~17 is two to three times that band — dirtier than almost any wealthy Western city, on a par with a mid-sized Chinese city after its decade of clean-up, and well short of the genuinely catastrophic: Delhi and Lahore spend much of the year north of 90.^[3] Not apocalyptic, then — but nowhere near clean, and squarely in the band where the health effects are well documented rather than theoretical.

That’s the calm annual average. On its worst days Johannesburg competes with anyone. On 31 May 2026, for a few hours, IQAir ranked it the most polluted major city on Earth — ahead of Delhi, Lahore and Beijing — at an AQI of 170.^[4] A city of jacarandas and swimming pools, top of the global filth table, on a cold winter morning.

And this isn’t one quirky backyard sensor: IQAir’s city-wide figure for Johannesburg in 2025 was 17.1 µg/m³.^[4:1] My single monitor, in a suburb people move to for the air, landed within a whisker of the entire city’s average. (I’ve also been pulling live readings from stations on other continents to place Johannesburg properly on this spectrum — a comparison that’s a whole post of its own, and it’s coming.)

🔗 Legal — but only just, and only on a technicality

The reason the number is so easy to wave away is that, on paper, my air is mostly legal. South Africa’s annual limit is 20 µg/m³,^[5] and my 17.4 slips under it. But “legal” turns out to be a slipperier idea than it sounds, and the law is doing something quietly generous.

Take the daily standard. It isn’t “never exceed 40 µg/m³.” It’s “don’t exceed 40 on more than four days a year.”^[6] That four-day grace is deliberate, and not unreasonable: one freak day — a nearby veld fire, a dead-still winter night — shouldn’t brand a whole area as non-compliant, so the rule is built to catch chronic pollution rather than one bad Tuesday. (Other countries do the same, often more generously: the European Union lets its daily PM10 coarse-dust limit of 50 µg/m³ be exceeded on up to 35 days a year before it counts as a breach.^[6:1])

Except my leafy-suburb station cleared 40 on five days — one past the allowance — and they cluster tellingly. Every one fell in the depths of winter: 19, 30 and 31 May, then 10 and 26 June 2026. The worst was 53.5 µg/m³ on 31 May — the very morning IQAir ranked Johannesburg the most polluted major city on Earth. So even measured against South Africa’s own lenient daily bar, my air doesn’t quite pass; it fails the WHO comprehensively and scrapes the local standard by a single winter day.

And that was the gentler ledger. The allowance runs per calendar year, my measurement year closed at the end of June — and July has spent its first two weeks demolishing what was left of the technicality. Four more days over 40 already (6, 7, 11 and 12 July, the worst at 48.5 µg/m³), which puts calendar-2026 at nine days over the daily limit — more than double the legal allowance — with half of winter still to come. The five-day near-miss you see in the chart below is last year’s story; this winter isn’t even pretending to pass.

Which raises the real question: why is the local bar set so low to begin with? It looks arbitrary, but it isn’t quite. South Africa’s 2012 standards were set with reference to the World Health Organization’s interim targets — a ladder of stepping-stone levels the WHO publishes specifically for heavily polluted, industrialising countries that can’t leap straight to the ideal.^[7] Our annual 20 sits between the second and third rungs of that ladder. It was set not at what’s healthy, but at what was thought achievable for an economy built on coal — which means the bar was, to a real degree, fitted to the pollution that already existed rather than to the limits of a human lung.

That gap is meant to close, slowly. On 1 January 2030 the South African annual standard tightens to 15 µg/m³ and the daily to 25.^[7:1] Even then it will stay looser than most of the wealthy world, and far above the WHO’s 5. A decade from now, we’re aiming at a bar the rich world has already stepped over.

🔗 So who decides what counts as “safe”?

The honest objection to all of this is: aren’t these limits just numbers somebody picked? Partly, yes — and it’s worth seeing how much they vary. Here is where different governments have drawn the line for the same pollutant, the same human lung:

Jurisdiction	Annual PM2.5 limit (µg/m³)
WHO guideline (health-based)	5
Australia	8
Canada	8.8
United States	9
European Union (from 2030)	10
Japan · South Korea	15
South Africa (now → 2030)	20 → 15
China (general zones)	35
India	40

Read it top to bottom and the pattern is unmistakable: the limit a country sets tracks its wealth and its industrial history far more than any difference in biology.^[8] Clean, rich Australia and Canada sit near 8; the US at 9; the EU is only now legislating its way down to 10 by 2030 (its standard sat at 25 for years); China is at 35, India at 40. Same particle, same lungs, wildly different lines. So the objection is half right: the legal limits really are negotiated compromises. They answer “what can we plausibly achieve and enforce?” far more than “what is actually safe?”

But look at the one number every government’s line sits above: the WHO’s 5 — and that one is not a political compromise. It is the output of the most rigorous process in the field. For the 2021 guidelines the WHO commissioned systematic reviews from five independent research teams and had an expert panel grade every finding with GRADE, the same evidence-rating method used for clinical guidelines.^[9] The heaviest evidence came from long-term cohort studies — millions of people tracked for years — linking annual PM2.5 to death from heart disease, stroke and lung disease.

And here is the part that reframes the whole question of what “safe” means: the studies keep finding harm at lower and lower concentrations, and have never found a floor. No threshold has been identified below which PM2.5 stops hurting people. So the WHO’s 5 is not a line between “harmful” and “harmless” — nothing that clean reliably exists in the data. It is roughly the lowest level at which the evidence is still strong, and about where the cleanest inhabited places on Earth actually sit. Below it, the harm very likely continues; we simply run out of clean-enough places to measure it precisely.

Which flips the “arbitrary” charge on its head. The laws are the arbitrary part — movable lines drawn where each economy can reach them. The science is the opposite of arbitrary: it points one direction, with no safe stopping point. “My air is legal” leans on the movable number and quietly ignores the fixed one. Measured against the number that isn’t up for negotiation, my leafy suburb — 3.5× it across the year, and more than seven times this winter — isn’t remotely close.

🔗 Why the trees don’t save you

This is the part that genuinely surprised me, because it runs against a deep intuition. We assume greenery is clean air — that a suburb thick with old trees and gardens must be breathing something better than the concrete parts of town. It’s half the reason people pay a premium to live under the canopy. But the canopy is a marker of wealth, not a filter, and the two have almost nothing to do with each other.

The reason is simple once you see it: air doesn’t respect suburb boundaries. A garden wall stops burglars, not particulates. The air over my leafy street is the same air mass that was over the mine dumps an hour ago and the highway before that — and a row of old plane trees does essentially nothing to the fine particles suspended in it. Street trees can catch some coarse dust on their leaves, but PM2.5 is far too small and too well-mixed for that to matter; if anything, on a still night a dense canopy traps ground-level air rather than scrubbing it. The greenery you can see is simply the wrong scale for the pollution you can’t.

Where does it come from? For Johannesburg the best breakdown puts roughly 37% of PM2.5 on industry and power, 17% on wood-and-coal burning, and about 7% on transport, the rest from dust and veld fires.^[10] Most of it is regional. Johannesburg sits on the Highveld, and to our east lies one of the densest clusters of coal-fired power stations on the planet. That air blows over us. (Strictly, the officially declared hotspot — the Highveld Priority Area — covers Mpumalanga and the East Rand, not the City of Johannesburg itself;^[11] but we share its sky and its weather.)

And then winter shuts the lid. Those cold, still nights are a temperature inversion: a layer of warm air settles above the cold air near the ground and clamps down, trapping everything we emit — every coal fire, every idling engine — in the few hundred metres we’re all breathing.^[4:2] Add the winter surge in burning for warmth, and you get both the doubled readings and the brown haze against the morning sun. My single most extreme minute of the year hit 771 µg/m³ — fifty times the WHO daily guideline — at 8:53pm on 20 November 2025, almost certainly a fire close by (the next-worst spikes, north of 500, came on another spring evening in September). Note the hour in both cases: after dark. The air can turn from fine to hazardous in the time it takes a neighbour to light a braai.

You can see the whole shape of it in a single view — every day of my record, coloured by how dirty the air was.^[12] The winters glow; the summers cool off; the gap in autumn is my station relocating.

Those evening spikes aren’t a coincidence, and the seasons aren’t the only cycle at work. Zoom all the way in — to the average shape of a single day — and the air keeps its own schedule: cleanest in the early afternoon, dirtiest after dark.

It’s the same physics in miniature: the midday sun stirs the air and clears it, and by evening the inversion drops the lid again — just as the dinner fires get going. Which hour you’re outside in, especially if you exercise, turns out to matter more than almost anyone realises. That’s a rabbit hole I’ll go down properly later in the series, when to run; for now, just notice that the air is never really still.

🔗 The part that should make us uncomfortable

It would be easy to write a self-pitying post here — poor me, breathing slightly imperfect air in my nice suburb. That’s not the point. The point is the inversion of it.

If a monitored, wealthy, tree-lined suburb averages 17.4, what is the air doing where nobody is measuring at all?

The burden isn’t shared evenly. An estimated 90% of households in Johannesburg’s socio-economically deprived areas — including informal settlements — still burn coal, wood and other polluting fuels to cook and keep warm.^[10:1] The people with the fewest resources breathe the same regional pollution I do, plus a thick extra layer of it, indoors, at close range, every winter night. The city’s exposure map is more or less its inequality map, drawn in smoke.

And here’s the catch that keeps the problem invisible: those same areas have the thinnest monitoring. South Africa’s official network exists — there are government stations in Soweto, on the East Rand, out in the Vaal — but it’s sparse, ageing, often offline for months, and reports at the level of a whole district rather than a street or a school. Private stations like mine are rarer still and cluster where the money is. So the communities harmed most have the least granular evidence of the harm — which makes it easy to underfund, easy to litigate slowly, easy to look away from. Not “no data.” Just never enough, and never close enough to where people actually breathe.

It took the courts to force even the coarse version of the issue. In the Deadly Air case, the groups groundWork and the Vukani Environmental Justice Movement took the government to court over the Highveld’s air. On 18 March 2022 the Pretoria High Court ruled that the pollution breached residents’ constitutional right, under section 24(a), to an environment that is not harmful to their health — believed to be the first time an African court had found a government in breach over air pollution.^[13] The government appealed; on 11 April 2025 the Supreme Court of Appeal dismissed the appeal and upheld the ruling.^[14]

But a court win is not clean air, and the aftermath is a lesson in how slowly this moves. The Highveld plan had existed since 2012; the regulations meant to give it teeth were drafted and then simply not published, year after year. It took the litigation to shake them loose — the Minister finally gazetted them only days before the appeal was heard — and the Supreme Court was scathing about the delay, calling it a lost opportunity and setting a hard deadline of 11 April 2026 to put binding, enforceable regulations in place.^[15] The second-generation Highveld plan now on paper aims for a 40% cut in emissions by 2030.^[15:1]

So the machinery is, at last, grinding. But look at what the “win” actually bought: a deadline to write rules, a plan with a target, a legal duty finally admitted. Not, yet, a single measurably cleaner breath for the people living under the plume — the enforcement against Eskom’s power stations and Sasol’s refinery, the region’s biggest emitters, is the unfinished war. The right to breathable air is now legally established in South Africa. What’s missing is the will to enforce it — and the fine-grained data to enforce it with, everywhere the monitors aren’t.

🔗 This is not just a number

Fine-particle pollution kills on a scale most people badly underestimate. The WHO attributes 4.2 million premature deaths a year worldwide to outdoor air pollution — most of them heart disease and stroke, not the lung conditions you’d expect; about 68% are cardiovascular.^[2:1] For Johannesburg alone, one estimate put the toll near 5,300 premature deaths in a single year.^[10:2] This is a leading cause of death hiding behind more visible ones.

That’s the weight behind my tidy 17.4. Not a statistic — a slow, quiet pressure on every heart in the city, mine included, and a heavier one on hearts and lungs across town that no sensor is watching.

🔗 Where this goes next

I started with a monitor and a suspicion that the “clean suburb” story was too comforting to be true. A year of data settled it. The air in one of the greenest, most privileged pockets of Johannesburg is legal, invisible, and — for half the year — several times worse than the world’s health authority says it should be.

If that’s the good part of the city, the parts we don’t measure deserve far more attention than they get. That’s the thread for the rest of this series — and the next post is the one that turned this from a hobby into something I can’t put down. I spent a year proving the air here isn’t clean. Next I went looking for what our children are breathing two suburbs over, where nobody is counting. I don’t think we’re going to like the answer.

---

All figures above from my own IQAir station (Jul 2025–Jun 2026, 310 days of data) unless cited otherwise. The chart is live from that dataset; guideline values and health statistics are sourced below and were verified in July 2026.

---

1. World Health Organization, WHO global air quality guidelines: particulate matter (PM2.5 and PM10), ozone, nitrogen dioxide, sulfur dioxide and carbon monoxide (2021). PM2.5 annual guideline 5 µg/m³, 24-hour guideline 15 µg/m³. https://apps.who.int/iris/handle/10665/345329 ↩︎

2. World Health Organization, “Ambient (outdoor) air pollution” fact sheet — an estimated 4.2 million premature deaths per year (2019) from ambient air pollution, of which ~68% are from ischaemic heart disease and stroke. https://www.who.int/news-room/fact-sheets/detail/ambient-(outdoor)-air-quality-and-health ↩︎ ↩︎

3. Comparative city annual-average PM2.5 figures from IQAir’s World Air Quality Report series (approximate recent annual means): London ~7.8 and New York ~7.5 (2024; New York’s 2023 figure was wildfire-inflated to ~11.6); Sydney roughly 4–6 (Australia’s national average was 4.5 in 2024); Beijing ~26–33; Delhi and Lahore frequently above 90 µg/m³. Values vary year to year; treat as order-of-magnitude context. https://www.iqair.com/world-air-quality-report ↩︎

4. IQAir, “Johannesburg among top 10 most polluted cities in the world” (31 May 2026): AQI 170 (“unhealthy”) and ranked the most polluted major city in the world that hour during an acute winter pollution episode; Johannesburg’s 2025 annual average PM2.5 reported as 17.1 µg/m³ (IQAir’s reported figure for 2024 was substantially higher, ~24 µg/m³ — city-wide averages swing year to year with winter weather and station coverage). IQAir also describes the Highveld topography trapping pollution under calm winter conditions. https://www.iqair.com/as/newsroom/johannesburg-among-top-10-most-polluted-cities-in-the-world-5-31-2026 ↩︎ ↩︎ ↩︎

5. South African National Ambient Air Quality Standard for PM2.5, gazetted under the National Environmental Management: Air Quality Act (Act 39 of 2004): annual average 20 µg/m³, 24-hour average 40 µg/m³ (with up to 4 permitted exceedances per year). https://www.gov.za/documents/national-environmental-management-air-quality-act-national-ambient-air-quality-standard-0 ↩︎

6. The 24-hour PM2.5 standard carries a “frequency of exceedance” allowance: the 40 µg/m³ daily limit may be exceeded on no more than 4 days per calendar year before the standard is breached. Such allowances are standard international practice — under EU Directive 2008/50/EC the daily PM10 limit value of 50 µg/m³ may currently be exceeded up to 35 times per calendar year (tightening to 45 µg/m³ with 18 exceedances from 2030; a 24-hour PM2.5 limit was itself only introduced in the EU’s 2024 revision) — and are designed to target chronic pollution rather than isolated events. South African Air Quality Information System, “Air Quality Standards and Objectives.” https://saaqis.environment.gov.za/Pagesfiles/Chapter 3 Air Quality Standards and Objectives.pdf ↩︎ ↩︎

7. South Africa’s National Ambient Air Quality Standards (2012) were set with reference to the WHO’s interim targets — graduated stepping-stone levels the WHO publishes for high-pollution jurisdictions working toward the guideline. The PM2.5 standards tighten from annual 20 / 24-hour 40 µg/m³ to annual 15 / 24-hour 25 on 1 January 2030. Source: SAAQIS “Air Quality Standards and Objectives.” https://saaqis.environment.gov.za/Pagesfiles/Chapter 3 Air Quality Standards and Objectives.pdf ↩︎ ↩︎

8. Annual mean PM2.5 limits by jurisdiction (µg/m³): WHO guideline 5; Australia 8 (a further goal of 7, slated for 2025, remains non-binding); Canada 8.8 (Canadian Ambient Air Quality Standard); United States 9 (EPA, 2024 revision, down from 12); European Union 10 (effective 2030 under the 2024 revised Ambient Air Quality Directive; 25 under the previous 2008 directive); Japan and South Korea 15; South Africa 20, tightening to 15 in 2030; China 35 for general (Grade II) areas, 15 for Grade I; India 40. Sources: Air Pollution & Climate Secretariat, “Air quality standards worldwide” https://www.airclim.org/working-areas/air-pollution/policy-initiatives/air-quality-standards-worldwide; Smart Air, “Global Air Quality Standards of PM2.5 and PM10” https://smartairfilters.com/en/blog/global-air-quality-standards-pm2-5-pm10/; IQAir 2024 World Air Quality Report. ↩︎

9. The WHO 2021 Global Air Quality Guidelines were derived from systematic reviews commissioned from five independent research groups, with a guideline development group grading the evidence using the GRADE method; guideline levels are defined as the lowest levels of exposure shown to harm health, drawing chiefly on long-term cohort studies of mortality. No safe threshold below which PM2.5 causes no harm has been identified. Sources: WHO, Global Air Quality Guidelines (2021), Executive Summary; WHO, “WHO global air quality guidelines” Q&A. https://www.who.int/news-room/questions-and-answers/item/who-global-air-quality-guidelines ↩︎

10. Clean Air Fund, Johannesburg city profile — PM2.5 source apportionment (industry and power ~37%, biomass burning ~17%, transport ~7%); ~5,300 premature deaths attributed to air pollution in Johannesburg (2019); ~90% of households in socio-economically deprived areas rely on coal, wood and other polluting fuels. https://www.cleanairfund.org/clean-air-africas-cities/johannesburg/ ↩︎ ↩︎ ↩︎

11. The Highveld Priority Area was declared on 23 November 2007 under section 18 of the Air Quality Act. It covers portions of Gauteng and Mpumalanga (the City of Ekurhuleni metro, Lesedi Local Municipality from the Sedibeng district, and parts of the Gert Sibande and Nkangala districts — nine local municipalities in all) — roughly 31,106 km² and ~3.6 million people. The City of Johannesburg is not within the declared area. Highveld Priority Area Air Quality Management Plan, Department of Forestry, Fisheries and the Environment. ↩︎

12. A note on two numbers that look inconsistent: my headline annual average of 17.4 µg/m³ is for the defined 12-month year, July 2025–June 2026 (310 days with data). The calendar heatmap above spans a slightly longer window — late June 2025 to mid-July 2026, about 327 days — which adds extra deep-winter days at both ends (including a July 2026 running above 36 µg/m³), so its average comes out marginally higher, at 18.0 µg/m³. Same station and same data pipeline; only the date window differs. ↩︎

13. Trustees for the time being of the groundWork Trust and Another v Minister of Environmental Affairs and Others (“Deadly Air”), Pretoria High Court, judgment delivered 18 March 2022 by Judge Colleen Collis; found a breach of the section 24(a) constitutional right to a healthy environment in the Highveld Priority Area. Brought by groundWork and the Vukani Environmental Justice Movement, represented by the Centre for Environmental Rights. https://earthjustice.org/press/2022/major-court-victory-for-communities-fighting-air-pollution-in-south-africa ↩︎

14. Supreme Court of Appeal dismissed the Minister’s appeal in the Deadly Air case on 11 April 2025 (Minister of Environmental Affairs v Trustees for the time being of GroundWork Trust and Others [2025] ZASCA 43), upholding the 2022 High Court judgment. https://groundwork.org.za/deadly-air-case-ministers-deadlyair-appeal-dismissed-by-supreme-court-of-appeal/ ↩︎

15. The SCA found that section 20 of the Air Quality Act imposes an obligation (not a discretion) on the Minister to make regulations implementing and enforcing the Highveld Priority Area Air Quality Management Plan; it criticised the decade-plus delay in doing so as unconstitutional and “a lost opportunity,” and ordered enforceable regulations within 12 months (by 11 April 2026). The Minister had gazetted draft regulations shortly before the appeal hearing; these went out for public consultation. The Second-Generation Highveld Priority Area AQMP targets an approximately 40% reduction in emissions by 2030. Sources: Daily Maverick, “Highveld deadly air: Supreme Court scathing of state’s regulations delay” (13 April 2025) https://www.dailymaverick.co.za/article/2025-04-13-highveld-deadly-air-supreme-court-scathing-of-states-regulations-delay/; Centre for Environmental Rights, Deadly Air case materials https://cer.org.za/. ↩︎ ↩︎