Indoor CO2 explained: the number that tracks ventilation

The number tracks how much outdoor air is reaching a room, weighed against the headcount inside it. Occupants breathe carbon dioxide out. Outdoor air dilutes it. The reading settles wherever those two rates balance, and it climbs whenever the second one falls behind.

NEA states this plainly in its own words. Section 5 of the April 2026 guidance on ventilation and indoor air quality is headed carbon dioxide levels as a proxy for ventilation. The text asks for a ventilation rate measurement where one is possible, and offers readings taken in occupied areas as the substitute where it is not.

The framing has sat in Singapore guidance for thirty years. It appears in a 1996 document on good indoor air quality for offices, prepared at the Ministry of the Environment by its Institute of Environmental Epidemiology. About 5% of exhaled air is carbon dioxide. Levels therefore rise in occupied rooms that are ventilated poorly, and the level is often used to judge how well ventilation is working.

Two properties make it a good proxy. The source is the occupants themselves, so the signal appears exactly when a room is fullest. And the gas mixes through the space instead of settling, so a reading taken at a desk describes the air people are breathing.

What it cannot describe is everything else in that air. Dust, cooking smoke, solvent vapour from a renovation and mould spores all move independently of it. A low number means the air is being replaced. It carries no claim about how clean that air was.

Three Singapore documents put a carbon dioxide figure on an occupied indoor space. They are three different kinds of instrument. A statute, a code of practice and an advisory guideline carry different weight. Quoting one while meaning another is the most common error in this subject.

The only statutory figure sits in workplace safety law. The First Schedule to the Workplace Safety and Health (General Provisions) Regulations lists permissible exposure levels for toxic substances. Carbon dioxide appears there at 5,000 ppm long term, measured across an eight-hour working day, and 30,000 ppm short term across fifteen minutes. Regulation 40 places the duty to stay under it on the occupier of the workplace.

Read that figure for what it is. A toxicity limit, listed among solvents and metals, sitting more than four times above the level an indoor air quality code advises for an office. No complaint about a stuffy meeting room is a complaint about approaching it.

The figure people should be quoting for an air-conditioned office comes from a code of practice. SS 554:2026, published by Enterprise Singapore, is the code that deals with indoor air quality. NEA reports that the standard recommends indoor carbon dioxide of not more than 700 ppm in excess of outdoor levels. With ambient outdoor air at about 400 ppm, NEA puts the resulting limit at approximately 1,100 ppm.

Its reach stops short of the home. The scope covers enclosed air-conditioned premises, then excludes residential premises outright, along with factory production areas, hospitals, polyclinics and laboratories. A flat sits outside it. Quote this figure at a homeowner and you have reached for a standard that does not cover their flat.

Two things about that figure deserve attention. It is a recommendation inside a voluntary standard, so it binds nobody by itself, though a regulator can make a Singapore Standard mandatory and a lease or contract that cites one turns it into a commercial obligation. And it is stated as a difference. The standard asks what the building added to the air it drew in, which is the honest way to ask whether ventilation is keeping pace with the people.

The familiar 1,000 ppm has a genuine Singapore source, and it is old. The 1996 office guidelines set it as a guideline maximum over an eight-hour average, beside carbon monoxide, formaldehyde and ozone. That document also says its guidelines carry a wide margin of safety. Occasional exceedance, on its own wording, is unlikely to produce toxic effects.

An absolute number ages in a way a difference does not. That 1996 document described unpolluted outdoor air as about 0.03%, which is 300 ppm. NEA's 2026 guidance puts ambient outdoor carbon dioxide at about 400 ppm. The outdoor baseline moved by roughly a hundred parts per million while the fixed indoor figure stayed exactly where it was.

One further number circulates and it has a narrow purpose. NEA suggests aiming below 800 ppm over the measurement period during an epidemic, following the CDC and REHVA. That is an infection-control target, and the same guidance is careful with it. High readings do indicate poor ventilation or overcrowding. The guidance still notes that levels may not track exposure risk from infectious pathogens directly.

A workplace can be perfectly cool and read badly at the same time. Nothing in the cooling loop touches the number, so the two measurements drift apart without anything breaking or sounding wrong.

Start with the box overhead. A split, a cassette or a fan coil pulls room air in, cools it and hands it straight back, so a split system does not ventilate at any fan speed. That mechanism gets its own treatment elsewhere. What counts here is the consequence: the equipment most premises actually have contributes nothing to the number on the display.

Outdoor air was somebody's design decision, taken well before the tenancy began. A larger building supplies it through a pre-treated air unit or an air handling unit, ducted out to an intake grille in the building skin. A strata shop unit frequently has no provision at all. Standing in the room reveals neither arrangement, which is why a first reading surprises people.

Headcount moves and the provision stays put. A floor drawn around one occupancy now holds another, and the outdoor air arriving per person falls in proportion. Meeting rooms show it first and worst. Small volume, high density, door shut, and the reading climbs steadily through the hour.

Timing explains a share of bad readings on its own. Base building ventilation plant runs to a schedule, so an early start or a weekend shift can put people in a room before the outdoor air supply comes on. Cooling is often available in hours when ventilation is not.

Conditioning outdoor air costs money, and that pressure never lets up. Air drawn from a Singapore afternoon arrives hot and wet, and the plant pays in electricity to cool and dry it before anyone benefits. Trimming that supply lowers the bill and raises the reading. Of everything on this list, that one was a deliberate choice.

The WSH Council and MOM guidelines on managing indoor air quality name three causes when a carbon dioxide level exceeds the limit. Over-crowding is the first. Insufficient outdoor air intake or air distribution is the second. Combustion activity in the space is the third.

Distribution is the one people forget. Air can arrive on a floor in the right quantity and never reach the corner the complaint came from. NEA's guidance addresses this directly, suggesting readings be used to find pockets of under-ventilated space and overcrowding, which means walking a meter around the floor instead of trusting one fixed point.

A carbon dioxide figure is weak evidence about cooling equipment in either direction. A high reading does not show the unit is faulty. A low one is no evidence that it is healthy. The two questions run on separate tracks and answer to different trades.

Nothing in the number reports on coil condition, refrigerant charge, drain fall or filter loading. Those faults announce themselves through other channels. A smell on startup, airflow that weakened across months, water where none belongs, and noise that changed character point at the unit.

Watch what gets quoted on the back of a bad reading. A chemical wash cleans the coil and lets the recirculation loop move more freely, and it adds no outdoor air whatsoever. A larger unit adds cooling capacity to a room whose shortfall was never about heat in the first place. If either gets proposed as the answer to a carbon dioxide reading, ask which part of the work brings outdoor air into the room.

An air cleaner meets the same test. Filtration targets particles, and NEA's guidance recommends filter grades for exactly that job. Carbon dioxide is a gas the occupants keep producing, and a filter standing in the room cannot remove it. A purifier may still earn its place on the floor. The question it answers was never the one the meter asked.

One overlap between the two trades is real and worth naming. Where a building genuinely supplies outdoor air, that air passes through cooling plant on its way in, so plant that is down or throttled takes the ventilation down with it. In that case the reading and the equipment are connected. Establishing whether the premises has such a provision comes first, because it decides whether there is anything for a contractor to look at.

Take the reading properly before spending anything on the strength of it. NEA's guidance describes a quick assessment as readings over at least five minutes per sampling location, taken in occupied areas at breathing height. A single spot value from an empty room at eight in the morning settles almost nothing.

Where the crowd changes through the day, NEA offers two routes. Install sensors with visible displays for constant monitoring, or take spot measurements during high occupancy. The peak is the condition worth knowing. An average smoothed across an empty afternoon buries it.

Measure the outdoor air too, because the code's figure is a difference. A reading of 1,100 ppm means one thing where outdoor air sits at 400, and something else where an intake overlooks a loading bay or a taxi queue. The 1996 guidelines built this in already, asking for at least two outdoor samples at the building entrance or at the fresh air intake.

Inexpensive meters are now everywhere, and the standard has caught up with them. SS 554:2026 added an informative annex on low-cost sensors, covering how to use them and how to interpret the data they produce. Treat a cheap display as a trend instrument. It is strong for showing that a room climbs through a long meeting, and weaker as a figure to put in a report.

NEA names two levers when a reading needs to come down, and both sit with the business. Increase the ventilation. Reduce the occupancy. Neither is a purchase from a cooling contractor, and the second one costs nothing beyond a scheduling decision about who sits where and when.

A formal route exists once the informal one runs out. Guidelines published by the WSH Council with MOM describe an indoor air quality management programme, with a named manager and a three-yearly audit. Further audits are triggered by occupant feedback, or by work done to close a previous finding. A meter reading is the cheap first pass. An audit by a competent person produces a document a landlord will act on.

Those guidelines carry one definition worth an operator knowing. Acceptable indoor air quality means a substantial majority of occupants, put at 80%, express no dissatisfaction with the air. The meter supports that judgement. It cannot substitute for it. Where readings look fine and people still complain, the room is saying something the meter was never asked.

• •Which document and which edition produced any figure quoted at you, since the 1996 office guideline and the current code disagree
• •What the outdoor reading was at the same moment, because the code states its limit as a difference
• •What the headcount was in the room when the reading was taken, and whether the door stayed shut
• •Whether any duct serving the space runs to a grille or louvre facing outdoors, or whether cooling is the whole of it
• •Who holds the operating schedule for that provision, if one exists, and which hours it actually runs