Why Two Meat Thermometers Disagree by 40 Degrees

This page contains affiliate links. If you buy through one we may earn a commission, at no cost to you.

Usually neither is wrong, and that is the useful answer. Two correctly calibrated thermometers can differ by forty degrees in the same piece of meat at the same moment, because they are not measuring the same thing. One is reporting where its sensor is. The other is reporting where its sensor is. The sensors are in different places, and in cooking meat different places are at wildly different temperatures.

Three separate mechanisms do this, and all three are checkable with numbers off the listing. General cooking information only here: for safe internal temperatures, follow current official food safety guidance rather than anything on this page.

First, the gradient, because it is enormous

A roast in a 350 degree oven has a crust close to 300 degrees because that is what browning requires, and a centre that might be at 155. That is nearly 145 degrees of difference, and in a three inch joint it is packed into an inch and a half of radius.

Average that out and you get close to 100 degrees per inch. So being half an inch out with your tip is worth tens of degrees, which is an order of magnitude more than the two to four degrees of calibration error a decent instrument carries. Arguing about which thermometer is accurate while being half an inch apart on insertion is straining at a gnat.

Second, the sensor is not at the tip

People insert to the depth that gets the pointy bit into the middle. The bit that reads is a bead or a junction some millimetres back up the stem, and no listing we have seen publishes where.

Worse, the stem is a metal rod with one end in 155 degree meat and the other in 400 degree air, so heat runs down it into the sensor. The standard rule in instrumentation is that this stem conduction error only becomes negligible once you have immersed something like ten to twenty stem diameters. That turns an abstract instruction into a measurement: a 4 millimetre probe wants 40 to 80 millimetres of steel inside the food, which is 1.6 to 3.1 inches. A thicker waterproof or wireless probe at 6 millimetres wants 2.4 to 4.7 inches.

Now compare that with what you are cooking. A chicken breast is maybe an inch and a quarter thick. A burger is three quarters of an inch. Neither can give a thick probe the immersion it needs through the top, which is why going in from the side along the length of the cut is not a tidiness preference. It is the only geometry that satisfies the number, and it is free.

And the clearest proof that the stem really is at two temperatures at once comes from a product that sells the fact as a feature. An InkGrill 4-probe wireless meat thermometer publishes four dual-sensor probes, each one reading food temperature from 14 to 212 degrees Fahrenheit and ambient temperature from 32 to 572 degrees at the same time, with Bluetooth to 300 feet, WiFi and a local mode, an app carrying 28 USDA presets, up to 25 hours of battery, IP67 dishwasher-safe probes and a magnetic anti-glare display.

Read that specification as a diagram. There are two sensors in one rod because the two ends of that rod are in different worlds, and the maker has put a number on each. The food sensor topping out at 212 degrees is the other half of the tell: it is not expected to see oven air, because it is meant to be buried. Its own record notes that full remote features need the app and home WiFi, that the ambient sensor is rated to 572 rather than to extreme heat, and that probes are the wear part of any thermometer. Our review of the InkGrill 4-probe thermometer sets out what is published.

Third, thickness buys waterproofing and costs you seconds

A probe reaching the temperature around it is an exponential approach, and the time constant scales with the diameter, because what matters is the ratio of the mass that has to be heated to the surface area heating it. A very fine junction settles in a second or two. Make the probe four times thicker to fit a waterproof seal, a battery or a second sensor inside, and the time constant goes up roughly four times with it.

Put a number on what that costs you if you read it early. Say the probe starts at room temperature, 70 degrees, and the meat is at 155, and the probe has a four second time constant, so it needs about 18 seconds to settle properly. Pull it at four seconds and it has closed 63 percent of the gap: it reads 127. At eight seconds, 148. At twelve seconds, 155 and a half. The whole of that 28 degree error at four seconds is an error of patience, and instant read is a claim about the format rather than about the physics.

Which is the right way to read a Presentastic wireless meat thermometer with Bluetooth and an app. Its record lists a fully cable-free probe in food-grade stainless with a high-temperature ceramic section, Bluetooth 5.2 at long range, app tracking with target alerts, a dishwasher-safe probe and a bamboo dock with a 2000mAh battery. A cable-free probe has to contain its own battery and radio, so it is necessarily a fat probe, and the ceramic section is the thermal break between the buried end and the end sitting in the oven. Both of those are consequences of the architecture rather than faults. The architecture also removes the response time problem entirely, because a probe you leave in has as long as it likes to settle. Its own record is clear about the trade: Bluetooth range drops through walls, there is no standalone display, and you have to charge it and respect the probe's temperature limits. More in our review of the Presentastic wireless thermometer.

The hardest case for all three mechanisms at once is a combination tool. A pair of KINGEAR smart BBQ tongs with a built-in thermometer publishes stainless steel tongs with an instant-read thermometer in the handle, a rechargeable lithium battery, a high-heat-resistant ergonomic grip, and the genuinely useful proposition that there is no small probe to lose. What its record does not publish is a probe length or a minimum immersion depth, which on the arithmetic above are the two figures that decide whether a reading means anything, and a sensing tip attached to a tool you are holding over a fire is the geometry least likely to get its ten to twenty diameters. Its own record also notes it is a spot check rather than a leave-in probe, that it needs charging and careful cleaning, and that in a combination tool one weak part affects the whole. See our review of the KINGEAR BBQ tongs. As a thing to hold while you turn food, it is fine. As the instrument you settle an argument with, it is the one that needs checking against a known point.

So which one do you believe

Neither, until you have made them measure the same thing. The test is two cups of water and it costs nothing.

Make an ice bath properly: more ice than water, stirred, and left a minute. That is 32 degrees Fahrenheit to within a fraction, anywhere on earth. Then do boiling water, and here is the correction almost nobody applies. Water boils at 212 degrees only at sea level. It drops by roughly 1.8 degrees for every thousand feet of elevation, so at 5,000 feet it boils at about 203, and a thermometer reading 203 in Denver is correct while one reading 212 is wrong by nine.

One trap in the test itself, which follows from everything above: you have to give the probe the same immersion in the cup that you would give it in the meat. Dip a 6 millimetre probe half an inch into a mug and the stem conduction error you are trying to measure is the error you have just created, and a good instrument will fail. Use a deep pan.

The honest limit

If both instruments pass both fixed points, then both are accurate and they were never the problem. The disagreement was position, depth and timing, and no purchase addresses any of the three.

What does address them is free. Insert along the length of a thin cut rather than down through it, so the stem gets its diameters. Push past the middle and then draw slowly back, watching the number fall and then rise, because the lowest reading you pass through is the centre and that technique finds it without guessing. Take the reading in at least two places on anything large or irregular, since the drawing above is a sphere and real joints are not. And wait. Watching the digits stop moving costs fifteen seconds and is worth more than every other sentence here.

Our existing simple guide to meat thermometers and safe cooking temperatures covers the targets themselves and the resting and carryover side, and our guide to grilling by temperature covers choosing between an instant read and a leave-in. This page is only about why the number on the screen may not be the number in the meat. For the targets, use current official food safety guidance.