How your Dexcom CGM reacts to calibration input

When you enter a calibration value into your Dexcom receiving device, your Dexcom data does not automatically line up on the new value.

it is easy to understand why. The measurements from a manual glucose monitor are only specified to be accurate with +/- 20% – an awfully poor expectation, but seemingly the only one that allows a reasonable cost for the strips used to read a BG value. On the other hand, your Dexcom, over time, averages out many calibrations (normally about two per day): after a while, it may actually have averaged out the measurement noise. In fact, some Dexcom users consider their Dexcom readings more trustworthy than a manual glucose meter reading.

Since the Dexcom algorithm averages out noise, it makes sense that it would not automatically line up your next reading to the last calibration.

Calibration higher than CGM

Experimentation shows that, if your calibration is higher than your Dexcom reading, in most cases the Dexcom algorithm will more or less average out its calculated new reading (taking into account your most recent BG change) with the new calibration value (modified by your recent BG change since calibration).

Calibration lower than CGM value

On the other hand, if the calibration value is lower than the last Dexcom reading, the Dexcom drift may potentially be more dangerous to you, since it now possibly overstates your BG and puts you at risk of a low. Experimentation shows that, when your calibration value is lower than your Dexcom reading, the Dexcom algorithm largely lines up on your last calibration (modified by your BG change since calibration).

What is the number Dexcom displays right after accepting your calibration input?

I am not quite sure what it represents – but it is NOT the calibrated number. The calibrated number is what will come out on the next regular tick after you have calibrated, whenever that is (within 5 minutes, obviously). The Dexcom algorithm corrects the calibration number by the drift it measures between the time you calibrated and the time of the next tick.

Calibration confirmation requested

In some cases, apparently when your calibration is far from the Dexcom algorithm expectation, your receiver device may actually ask you for a second calibration to confirm the first. These cases are rare.

In brief

So, as a simplification:

  • if your calibration is higher than the Dexcom measurement, the algorithm appears to average the two;

  • if your calibration is lower than the Dexcom measurement, the algorithm appears to adopt the new calibration as the new value.

Consequence: Dexcom values are normally lower than BG

Since the algorithm skews towards lower values (by going up half way when low, but all the way when high), your Dexcom generally reads lower than the measurement of your BG meter. Over the life of a sensor, the calibration pattern often shows regular readjustments upwards for the Dexcom values.

End of wiki ---------- comments start here


More space around the pic, above and below?

1 Like

[Mod comment: part of this post spawned another thread: Dexcom time lag, and interstitial fluid glucose vs BG]

So when doing calibration, and dexcom updates display value, what does it represent.?


This is not accurate for me—just tested it out since my Dex was 85, I felt lower, tested and was 71, and inputting that led to a Dex reading of 78. My Dex was not soliciting a calibration at that time, so not sure if that plays into it, but I’m pretty sure in general that hasn’t been my experience either.

Also, I find my 90-day Dex averages and A1cs to align perfectly every time so far, for what it’s worth.

Interesting. I have seen this happen a small number of times, but, in general, I was able to attribute it to a change in BG between the moment the value was inputted and the moment it is displayed on its regular tick.

And the Dex average always underestimates my A1c. Pbbth!

My next reading was a 68, with a slanted down arrow (previously a forward one), so take that for what you will. At the very least, I take it to mean my internal hypo sensors are working well ha. It does seem like the re-calibration worked, even though it didn’t function to exactly re-align with the inputted value.


@Cardamom, when you wrote that the Dexcom went to 78, was that the instant reading, or the one that came on the pulse after you entered the calibration?

Now that I read your last message, I am now understanding what happened as follows (each dot is a minute):

85CGM . 71BG/78Receiver . . 68CGM

Is that correct? if so, it would make sense to me, as I believe that the first value showing on the receiver is an interim value (not a full calibration value) – I think the full calibration only shows in the pulse that appears after calibration, not in the value that is displayed right after you enter your BG.

I’m using my phone, not the receiver, but yes, 78 was what it adjusted to immediately after I put in the calibration. The value following that was 68, and the arrow changed. Neither value was the exact one I entered. That said, I know if I do the double calibration trick, that immediate value is always exactly the one I enter.

Bear in mind, it is my suspicion that a “double calibration” may erase all previous calibrations. This is only a guess. But if so, this is potentially not helpful under most circumstances.

In the case where you think the previous calibrations were just not good which are resulting in bad readings and you would just as soon do a sensor stop/start but don’t want to wait the 2-hr blackout then perhaps the double calibration performs the same functionality from a calibration point of view.

Guess work. Based on the fact that the first calibration results in an average resulting from the cgm as shown on the receiver and the first calibration entered. However the second calibration does not perform the “average” approach again but rather immediately sets the receiver (or SmartApp - same difference) to exactly the calibration number.


OK, then it worked just as expected:

  • I am not sure how the 78 is computed, but it is an intermediate value that does not represent the calibration

  • I think that 68 = 71 + TimeTrendCorrection, where TimeTrendCorrection represents the change in BG detected by the Dexcom sensor between the moment you calibrated and the moment the next pulse is displaying. So that would mean that your BG went down by 3 in the minutes between calibration and the next pulse.

There may also be some correction of the calibration value in there. But, essentially, 68 is 71 your calibration corrected by a small factor (I guess the BG change since).

1 Like

Yes, I agree—I only do this if it’s really out of whack or on the rare occasion when I’ve switched sensors without restarting (which is possible sometimes, if you need to swap out a faulty sensor before your 7 days are up and don’t want a gap right then).

1 Like

Ok that makes sense!

1 Like

I clearly need to explain the process better (at least what I think it is) in the wiki post above. Will work on that.

@cardamom, I reformatted and reworded the wiki statement – can you review it and let me know if you think it is more clear and understandable from the point of view we discussed?

1 Like

I think so, yes!


@MM2, sorry for taking so long to reply – I had to research it in depth. I think these are two great subjects. So I am splitting the other subject (lag time and tissue vs blood glucose).

Calibration vs updated: the number that dexcom flashes right after calibration is not fully calibrated. It can be very different from the one that shows up on the tick that comes after calibration. It is that following tick number that embodies the calibration.

I am not 100% sure what the number is – it seems to be just a rough average of your calibration + the existing CGM number. But, as discussed above, this rough average can be very far from the final number (a) if your calibration is lower than the CGM and (b) if your BG is varying fast.

For instance, let’s assume that, when your calibration is lower than the CGM, Dexcom takes that calibration at face value. let’s say that you enter this Calibration on minute 1 after a Dexcom measurement tick. You have 4 minutes after your calibration and before the next tick, during which the Dexcom calculates that Variation. So, the value that displays on the tick will be fully calibrated as:

TickValue = Calibration + Variation

1 Like

I know I keep responding the long way around, but this sounds an AWFUL lot like what the Medtronic sensors do. I think you responded to me somewhere, but I’m not sure where, so pardon me if I’m repeating the same question. Is there any chance this is what we see with Medtronic?? And what would one do, say, if they wanted to tighten that calibration when one’s insulin delivery seems to be dependent on it?

I think what is not being mentioned in this thread is that Dexcom is not showing you your BG, it is showing you what your interstitial reading is at the time. I have spoken to some of the higher level Dexcom reps about their algorithm and what they are showing with the reading.

Dexcom expects your interstitial reading will be within 20 points or 20% of your BG number, but they do not immediately show your BG number when you calibrate because they are not showing you your BG, only your interstitial reading.

That’s why lying to the Dexcom when you calibrate is a bad idea. If you do not enter a correct number, it associates the incorrect number with the interstitial reading at that time, and that can screw it up even worse.

According to my talks with Dexcom, there is no “prediction” about your current BG. Your Dexcom reading simply displays what it interprets your current interstitial reading to be.

Talk to 2nd or 3rd level Dex support and discuss it with them. It makes a lot more sense in the context of everything they are doing with calibrations.