It is 6:40 in the morning at a distribution yard outside Bhiwandi. A driver blows into the device, the screen settles on 0.02, and the supervisor holding it has no idea what to do next. The driver says he had nothing since the previous evening. The vehicle is loaded, the customer slot is at nine, and three people are now standing around a handset arguing about a two-digit number.
That argument is almost never about the device. It is about the fact that nobody decided, in advance, what an alcohol breath analyser level means in that yard and what happens at each band. The number is only useful if someone has already written down the response.
This article explains what the reading represents, how the units convert, what India's legal threshold actually is, and how to set a workplace cut-off that survives a dispute.
What the reading is measuring
A breath analyser does not measure alcohol in blood. It measures alcohol in deep lung air and converts that to an estimated blood alcohol concentration using a fixed ratio, conventionally around 2,100:1. Everything you see on the screen flows from that conversion.
That is why the same breath sample can be displayed in four different ways depending on the device and the market it was built for. None of them are wrong; they are the same quantity in different clothes.
How it is expressed Value at India's driving threshold
BAC as a percentage 0.03%
Milligrams per 100 ml of blood 30 mg/100 ml
Milligrams per litre of breath approximately 0.15 mg/L
Micrograms per 100 ml of breath approximately 15 µg/100 ml
If your devices and your policy document use different units, you have built a dispute into the system. Pick one unit, state it in the policy, and buy devices that display it.
The legal alcohol breath analyser level in India
Under Section 185 of the Motor Vehicles Act, the threshold for driving is 30 mg of alcohol per 100 ml of blood, detected by a breath analyser. In percentage terms that is 0.03% BAC, which is well below the 0.08% figure quoted in most American material you will find online.
This matters when you are reading international content. A page that tells you 0.05 is "under the limit" is not describing Indian roads. On an Indian highway, 0.05% is comfortably over.
The 2019 amendments raised the consequences for driving over the threshold, and repeat offences are treated more severely than first ones. For a commercial operator, the licensing and insurance fallout from a drink-driving case usually costs far more than the penalty itself.
Workplace thresholds are not road thresholds
Here is where most site policies go wrong. The 30 mg figure is a prosecution threshold for driving. It is not a statement that 29 mg is safe near a conveyor, a crane, a confined space entry or a live electrical panel.
Most serious Indian workplace policies in mining, oil and gas, ports, heavy manufacturing and aviation ground operations run a zero-tolerance standard instead. Flight crew, for example, are tested on a zero-tolerance basis before duty. The reasoning is simple: impairment of judgement and reaction time begins well below any statutory driving limit, and a site cannot run a different limit for every task.
Zero tolerance in practice means the policy declares a cut-off at or very near 0.000, and anything detectable triggers a defined process. It does not mean the supervisor improvises.
Choosing your own cut-off
Three questions settle it:
• What is the worst thing a slightly impaired person can do here? Drive a tanker, or sit at a desk?
• Can your device reliably resolve the level you are claiming to enforce? A cheap semiconductor sensor that is honest at 0.05% may be noisy at 0.01%. Claiming enforcement at a level your hardware cannot hold is worse than having no number at all.
• What is the second step? A cut-off without a stand-down procedure, a retest rule and an HR pathway is decoration.
Why two devices can show different levels for the same person
Supervisors lose confidence when the yard device and the gate device disagree. Usually there is a reason, and it is rarely mysterious.
Sensor type. Fuel cell sensors respond to ethanol specifically and hold their accuracy better at low readings. Semiconductor sensors are cheaper and can respond to other substances in breath, which is why a diabetic driver or someone who has just used a solvent can produce an odd result. If your policy enforces near-zero, a fuel cell device is not a luxury.
Calibration status. Every sensor drifts. A device that has not been calibrated for a year is producing numbers, not evidence. Calibration intervals and the availability of service are a bigger part of long-term cost than the purchase price, which is one of the themes in our breakdown of what decides the number on the quote.
Sampling technique. A short, weak blow may capture upper airway air rather than deep lung air and under-read. A device with active sampling and a flow sensor is harder to cheat than one that accepts any puff.
Mouth alcohol. Residual alcohol in the mouth from mouthwash, certain cough syrups, some mouth fresheners, a recent sip or even belching can produce a high reading that has nothing to do with blood alcohol. This is the single most common cause of a disputed first result.
The fifteen-minute rule
Wait fifteen to twenty minutes after the last drink, rinse, cigarette or burp, then retest. Mouth alcohol dissipates in that window; alcohol in the bloodstream does not. If the second reading matches the first, the result is real. If it collapses to zero, you avoided accusing someone on the strength of a mouthwash.
Build the retest into the policy so it is automatic rather than a favour. The mechanics of running a clean test are covered in more detail in our guide to the breath analyser test.
Turning a level into a decision
A workable policy maps bands to actions. The exact cut-offs are yours to set; the structure should look something like this.
Reading band Likely meaning Standard response
0.000 No detectable alcohol Clear for duty, log the result
Trace, below your cut-off Possible mouth alcohol or residual Wait 15 minutes, rinse, retest
At or above your cut-off, below 0.03% Policy breach, not a road offence Stand down from safety-critical work, HR process
0.03% and above At or over the statutory driving threshold No driving under any circumstance, escalate per policy
Two things make this hold up later. First, every test is logged, including the negatives, so the record shows a system rather than targeted checks. Second, the person tested sees their own reading. Screening that happens out of sight of the person being screened is the kind that collapses in a grievance hearing.
When a breath reading is not enough
Breath screening is fast, non-invasive and appropriate for gate and pre-shift use. It is a screening tool. Where the outcome is likely to be contested in a formal proceeding, a confirmatory blood test ordered through the proper medical channel carries more weight, and the difference between the two is worth understanding before you need it. We set the comparison out in blood alcohol tester vs breath analyser.
The practical arrangement most sites settle on: breath analyser at the gate for everyone, every day; medical confirmation reserved for the small number of cases that go further.
Matching the device to the level you enforce
If your policy enforces zero tolerance, the device must be credible at low readings, must display the unit your policy names, must be calibrated on a schedule with a certificate you can produce, and must be serviceable in India without shipping it overseas.
That narrows the field quickly. Our range of alcohol testing machines is grouped by use case, from handheld pre-shift screening units to higher-throughput devices for gates where a hundred people pass in twenty minutes. For sites deciding between models, the practical question is not peak accuracy on a spec sheet but accuracy after eighteen months of dust, heat and daily use.
The supplier relationship matters as much as the hardware, because calibration is a recurring commitment rather than a one-time purchase. Our note on how to choose a supplier that still answers the phone in year three covers what to ask before the first invoice.
The decision in front of you
The supervisor in the yard at 6:40 was not failing at his job. He was improvising because nobody above him had written down what 0.02 means at that gate.
Pick a unit. Pick a cut-off you can defend and your device can resolve. Write the retest rule, the stand-down step and the logging requirement. Then the number on the screen stops being an argument and starts being an instruction.