Every mileage conversation in India eventually lands on "kitna deti hai?", and almost everyone answers with the wrong number, pulled from the wrong place. The tank-to-tank method is the only calculation that cannot be faked by the vehicle's own electronics.
The instrument cluster gives you an average. The manufacturer gives you a test-cycle figure. Your neighbour gives you what he remembers from three months ago. None of these is your mileage. Your mileage is what your specific vehicle, running your specific routes, on the fuel in your specific city, produces right now. The only way to get that number is to measure it yourself, across a full tank of fuel, from brim to brim.
This is not a complicated procedure. But it has a few steps where people consistently go wrong, and those errors compound. Get them right once and you will have a baseline you can trust.
Why the Tank-to-Tank Method Works When Others Don't
Your dashboard's fuel economy display calculates consumption from injector pulse-width data, not from actual fuel movement. It is a model, not a measurement. The model has calibration tolerances, and it drifts further from reality under conditions the factory never tested: local ethanol blend variation, tyre pressure differences, a cold engine warming up on a short commute, air conditioning load on a 42°C afternoon in Nagpur. Errors of 5 to 15 percent between the displayed figure and the real figure are routine, as covered in more detail in the post on why dashboard averages lie and what the pump method actually measures.
The tank-to-tank method bypasses all of that. You are measuring the one thing that cannot lie: how many litres of fuel disappeared from the tank over a known distance. Divide distance by fuel, and you have kilometre-per-litre. No electronics involved.
The Exact Steps, Without the Common Errors
Step 1: Fill to the same point, twice
Go to a pump you use regularly. Fill the tank until the nozzle cuts off automatically. Do not top up after the auto-cutoff; that adds an inconsistent amount each time and ruins the baseline. Let the nozzle click off, pay, and drive away. Note nothing yet about fuel quantity. This is your start point, and the only thing it establishes is a physical level in the tank.
Reset your trip odometer to zero at this moment. Not at home, not later. Right at the pump, before you pull away.
Step 2: Drive a meaningful distance
The longer the interval, the more accurate the result. A single short commute of 15 km is too noisy: a 200 ml measurement error at the pump becomes a 13 km/l swing in your final number. For a car used mostly in city traffic, aim for at least 200 km between measurements. For a highway run or a longer trip, a single tank covering 350 to 500 km is ideal.
Drive normally. Do not change your habits for the measurement run: keep the AC running if you always run it, take the same roads, maintain your usual speed. You want to measure what your vehicle actually does, not a best-case lap. A number you manufactured by hypermiling tells you nothing useful.
Step 3: Return to the same pump, same nozzle if possible
Pump calibration varies slightly between machines at the same station and more between different stations. For your baseline measurement, return to the same pump. Fill again until the nozzle auto-cuts off, same as before. Do not top up. Read the litre figure from the receipt or the pump display before it resets. This is the fuel volume you consumed over the interval.
Write down two numbers: the litres dispensed, and the trip odometer reading.
Step 4: The arithmetic
Mileage (km/l) = trip distance (km) divided by litres dispensed.
An example with real numbers: you drove 347 km and the pump dispensed 28.4 litres at the refill. 347 divided by 28.4 gives you 12.2 km/l. That is your mileage for that tank, on those roads, in that weather, with that load.
If you drive a CNG vehicle, the unit changes to km/kg, since CNG is sold by weight. The method is identical: note the kg dispensed at refill, divide distance by kg. A mass-market hatchback running CNG in city traffic might show 18 to 22 km/kg depending on the city, the cylinder pressure at the start of the refill, and ambient temperature. A single reading is a data point; three or four readings averaged across different weeks is a number you can argue with.
Sources of Variance You Should Know About
Even a perfectly executed tank-to-tank measurement has some variation. Knowing which sources are real and which are noise keeps you from over-reading a single data point.
Tyre pressure. A tyre that is 4 PSI underinflated produces meaningfully higher rolling resistance. Over a 300 km tank, that can cost you 0.5 to 1 km/l. Check pressure cold, once a month, and note when you changed it so you can compare before and after readings.
Fuel load and passenger weight. A car carrying three adults and luggage on a Pune-to-Goa run will consume more than the same car with one person on a flat city loop. If you are trying to isolate the effect of a single variable (say, the E20 transition), keep the other variables consistent across the comparison tanks.
Route mix. A tank that is 80% highway driving will show better km/l than a tank that is 80% stop-and-go in Bengaluru traffic. Label your records: "city," "highway," or "mixed." A mixed-cycle average across several tanks tells you more than any single reading.
AC vs. no AC. In summer months, particularly in cities like Hyderabad, Chennai, or Ahmedabad, air conditioning can account for a visible drop in mileage. If you are doing a controlled comparison, note whether AC was on.
E20 transition effects. Since April 2026, all petrol at Indian pumps is E20. If your vehicle was calibrated for an earlier blend, you may see a modest mileage drop. The measured range in published test-agency data is roughly 2 to 6 percent higher fuel consumption compared to E10, with the effect smaller in newer, E20-tuned vehicles. The only way to see what E20 does in your specific vehicle is to log several tanks and compare against your earlier average. A vehicle bought before April 2023 should have its fuel-lid sticker and owner's manual checked for compatibility notes; see the guide on measuring E20's actual effect on your own mileage for the full methodology.
Building a Baseline Over Multiple Tanks
One tank-to-tank reading is useful. Four readings averaged is a baseline. Eight readings across different seasons is a real picture of your vehicle's behaviour.
What you are looking for in a multi-tank log is not the highest or lowest reading, but the central tendency and the spread. If your city-traffic readings cluster between 11.8 and 12.6 km/l across six tanks, your baseline is about 12.2. If one tank comes in at 10.1, that is a signal to investigate: did you run the AC constantly, drive into a headwind all day, sit in a two-hour jam on the Western Express Highway? If no obvious cause, check tyre pressure and whether the vehicle needs a service.
A meaningful shift in your baseline, say a consistent drop from 12.2 to 11.3 over three consecutive tanks with no change in route or load, is worth taking to a service centre. Your log is the evidence.
Bharometer
Log every fill-up in seconds, scan the pump screen to auto-fill litres and amount, and watch true tank-to-tank km/l build into a baseline you can actually trust.
Where Manual Logging Breaks Down (and How to Fix It)
The method above works perfectly if you actually record the numbers every single time. That is the failure point for most people: the receipt gets lost, the trip odometer gets reset for a different reason, the km reading is noted but the litres are forgotten. Three weeks later you have a partial log and cannot reconstruct the tank.
Bharometer is designed specifically around this problem. You log a fill-up by entering the litres dispensed and the amount paid; alternatively, the pump-scan feature lets you photograph the pump display and it auto-fills the litres, amount, and per-litre rate from the image. Odometer reading goes in, and the app calculates true tank-to-tank km/l immediately, using the distance since the previous logged fill-up. No manual division required.
Over time, your fill-up history becomes a chart: km/l by tank, spending by month, cost per kilometre. You can see the E20 transition in your own data if you have been logging since before April 2026. You can see what the monsoon does to your numbers in July versus what February highway trips produced. The log is the asset; the app keeps you from losing it.
For CNG users, Bharometer handles km/kg mileage directly, and it supports dual-fuel vehicles where you might run petrol to start the engine and switch to CNG once moving. Each fuel is logged and tracked separately, so your petrol-mode efficiency and CNG-mode efficiency stay distinct rather than blended into a meaningless combined figure. City-level daily fuel prices and nearby pump locations are also in the app, which helps when you are deciding whether to fill now or wait.
The data stays on your device and in iCloud. No account required, no ads. It is available on iPhone through the App Store; an Android version is also available free on Google Play.
What to Do With Your Mileage Number
Once you have a reliable baseline, you have a diagnostic tool. Compare city versus highway splits. See how a tyre pressure correction changed your numbers. Track whether a scheduled service improved your km/l. If you are evaluating the cost of switching from petrol to CNG, your actual ₹/km on both fuels is in the log, not in a manufacturer brochure. The post on what your vehicle actually costs per kilometre in India goes further into how to use these numbers for total running-cost budgeting.
Mileage is not a bragging number. It is a diagnostic number. The only version worth quoting is the one you measured yourself, tank to tank, on your actual roads, with your actual habits. Everything else is noise.