If 50/50 is good, 100% must be better. That is the logic behind every jug of concentrate poured straight into a radiator, and it is wrong in a way the numbers make very plain. Straight ethylene glycol protects worse than a 50/50 mix. A 60/40 mix, on the other hand, really does buy a lot more cold protection. Both facts sit in the same table.
All the freezing and boiling points below come from Dow's published product sheet for an ethylene glycol coolant base, by percentage of glycol in the water by volume. The amounts to drain and refill come from the Coolant Mix Calculator, worked out for a 3.5 gal cooling system. Your manual's capacity and the coolant type your vehicle maker specifies come first; this is about the arithmetic behind the ratio.
Why bother with a table at all, when the jug has a chart on the back? Because the jug chart starts at 50/50 and stops there. It tells you what a correct mix does, not what your mix does after three top-offs with water, a leaking hose and a season of guessing. A coolant tester and the numbers below close that gap.
Myth: straight antifreeze gives the best protection
No. Pure glycol is not the best antifreeze, water is part of the recipe. In Dow's table a mix at 95% glycol freezes at -3.0 °F. A plain 50/50 mix freezes at -35.1 °F. So the jug poured in neat protects about 32 °F worse than the mix printed on its own label.
Why does anyone do it? Because "more is better" sounds right, and because the jug says antifreeze, not "half of what your engine needs". Can you run straight antifreeze? You can pour it in. It just stops doing the job it was bought for.
Myth: every extra percent of glycol adds protection
Only up to a point, and then it turns around. Here is the whole curve, strong end included:
| Ethylene glycol, % by volume | Freezes at | Boils at (no pressure cap) |
|---|---|---|
| 37.5% | -8.1 °F | 222 °F |
| 47.6% | -28.9 °F | 225 °F |
| 52.7% | -42.0 °F | 227 °F |
| 57.8% | -54.9 °F | 230 °F |
| 62.8% to 73.6% | below -60 °F | 235 to 248 °F |
| 78.9% | -52.2 °F | 255 °F |
| 84.3% | -34.5 °F | 273 °F |
| 89.7% | -21.6 °F | 285 °F |
| 95.0% | -3.0 °F | 317 °F |
The best freeze protection in the sheet sits between about 63 and 74% glycol, where Dow only says "below -60 °F". Past that it climbs back up: -52.2 °F at 78.9%, -34.5 °F at 84.3%, roughly where 50/50 started. So a 70/30 mix still protects very well. A 90/10 one does not.
Myth: 60/40 runs a lot cooler in summer
The boiling point does go up, but slowly. At 50% the mix boils at 226 °F at normal air pressure, at 57.8% at 230 °F. That is 4 °F for almost eight more points of glycol. The pressure cap on a running engine raises the real boiling point well above both numbers, so the difference between a 50/50 and a 60/40 is small next to a cap that holds pressure properly.
And glycol is not free in summer either. The same Dow sheet shows water carrying heat better: at 50 °F a 40% mix holds about 5% less heat per pound than a 30% mix (specific heat 0.821 against 0.865 Btu/lb °F).
Myth: topping off with premix will strengthen a weak mix
A 50/50 premix can pull a mix toward 50%, never past it, and it gets there slowly. The coolant tester says 30%, you want 50%. With concentrate the Coolant Mix Calculator says: drain 1.00 gal from a 3.5 gal system and put back 1.00 gal of concentrate. With premix the answer is 3.50 gal, which is the whole system. In other words a full drain and refill.
| Full 3.5 gal system | Drain and refill with concentrate | Drain and refill with 50/50 premix |
|---|---|---|
| 20% to 40% | 0.88 gal | 2.33 gal |
| 25% to 40% | 0.70 gal | 2.10 gal |
| 30% to 40% | 0.50 gal | 1.75 gal |
| 30% to 50% | 1.00 gal | 3.50 gal (all of it) |
| 35% to 50% | 0.81 gal | 3.50 gal (all of it) |
For smaller corrections premix works, just with roughly three times the liquid. From 25 to 40% it is 2.10 gal of premix against 0.70 gal of concentrate. If the tester reads low, buy concentrate.
Myth: one quart of concentrate fixes a weak mix
It is the most common shortcut: the tester reads weak, so a quart of concentrate goes in after a quart of old coolant comes out. In a 3.5 gal system at 30% that quart (0.25 gal) moves the mix to 35%. The freeze point goes from 3.5 °F to -4.1 °F. Better, but nowhere near the -35.1 °F of a real 50/50, which takes 1.00 gal, four times as much.
The reason is simple dilution. Every quart that comes out already holds glycol, so each quart of concentrate that goes in adds less than its own volume of new glycol. The weaker the mix, the less this matters; the closer to the target, the more you have to swap.
A worked case: a 4.5 gal truck that needs -40 °F
Take a larger system, 4.5 gal, that tests at 35%, good to -4.1 °F. The owner wants protection to -40 °F. In temperature mode the calculator turns that into 52% glycol and says: drain 1.18 gal, refill with 1.18 gal of concentrate. The result freezes at -40.2 °F.
Try the same with premix and the calculator refuses, because 52% is above what a 50/50 product can reach. That refusal is the useful part. It saves a trip to the store for the wrong jug.
Fresh fill: how much concentrate for your system size
After a flush or a radiator replacement the system is empty and the question is simpler: how much of each. Here are four common capacities, for a plain 50/50 and for protection to -40 °F:
| System capacity | 50/50: concentrate + water | To -40 °F (52%): concentrate + water |
|---|---|---|
| 2.5 gal | 1.25 + 1.25 gal | 1.30 + 1.20 gal |
| 3.5 gal | 1.75 + 1.75 gal | 1.82 + 1.68 gal |
| 4.5 gal | 2.25 + 2.25 gal | 2.34 + 2.16 gal |
| 6 gal | 3.00 + 3.00 gal | 3.12 + 2.88 gal |
The two columns differ by only a few tenths of a gallon. That is the point of the table: going from "the usual" to "the coldest night in the upper Midwest" costs very little concentrate when you start from empty.
Myth: a mix that is too strong just needs water in the reservoir
What happens if there is too much antifreeze? A full cooling system has no room for the fix. Pouring water into the overflow tank only fills the tank. The correction is the same drain and refill as above, only with distilled water instead of concentrate.
A system that tests at 65% glycol needs 0.81 gal drained and replaced with water to come back to 50%. The freeze point goes from "below -55 °F" to -35.1 °F. The boiling point at normal pressure drops from about 238 °F to 226 °F.
Opening the radiator drain rarely empties the whole system. Coolant left in the block and the heater core mixes with the new fluid, so run the engine with the heater on, let it cool, and test again before the last top-off.
Myth: the label ratio is all you need to know
The ratio is a means to an end. The end is a freeze point a few degrees below the coldest morning you expect, and Dow suggests at least 5 °F of margin. Pick the temperature first and let the ratio follow:
- Protection to -20 °F takes about 43.7% glycol: 1.53 gal of concentrate and 1.97 gal of water in an empty 3.5 gal system.
- Protection to -40 °F takes about 52%: 1.82 gal of concentrate and 1.68 gal of water.
Both come straight out of the calculator's temperature mode. Compare them with what the tester reads before buying anything.
The one that holds up: 60/40 does buy real cold protection
Here the popular belief is right. Going from 50% to 57.8% glycol moves the freeze point from -35.1 °F to -54.9 °F, about 20 °F colder for less than eight extra points. In a full 3.5 gal system that is 0.55 gal drained and replaced with concentrate.
Is that worth it? In Minnesota, North Dakota or Alaska it can be. In Georgia it is money spent on protection the car will never use. The calculator makes the trade explicit: the same screen shows the freeze point, the boiling point and the gallons it takes.
What a refractometer is actually reading
A refractometer does not measure glycol directly. It measures how much the fluid bends light, and glycol raises that number in a steady way. Dow's sheet lists it next to every row: a refractive index of 1.3328 for plain water, 1.3635 for a mix that is 30% glycol by weight and 1.3849 at 50% by weight. A coolant refractometer turns that reading into a freeze point on its scale, which is why one drop is enough.
Floating-ball testers work from density instead. Either way, the reading is the "current %" the calculator asks for, and it beats guessing from the color of the fluid, which tells you about the dye and nothing about the strength.
Three numbers to check before winter
- The coolant type and ratio your vehicle maker specifies, from the owner's manual. That overrides every chart.
- The real strength in the system, with a refractometer or a coolant tester, not the color of the fluid.
- The coldest temperature you expect, minus 5 °F. Put that into the calculator's temperature mode and compare with the tester reading.
The calculators behind this article
- Coolant Mix Calculator - concentrate, premix and water for a fresh fill, or how much to drain and refill, with freezing and boiling points.
- Compression Ratio Calculator - static compression from bore, stroke, chamber cc, gasket and deck, for the day the engine itself comes apart.
- Bolt Torque Calculator - torque in ft-lb and N-m from bolt size, SAE grade or metric class and thread condition.
More for the garage
- Capacity Converter - quarts, gallons and liters for the capacity in your manual.
- Car Maintenance Schedule Calculator - when the next service is due, by date or by mileage.
- Oil Life Calculator - how much life is left in the engine oil.
- EV Winter Range Calculator - what cold weather does to an electric car instead.
- Car Service Cost Calculator - a dealer against an independent shop over a year.