I'm setting up the computer and looking at the manual's chart for a 700x25, the number is 2096. However, the measured circumference of the tire is 2135 which happens to be the manual's number for a 28 tire.

So, slightly confused, I set the computer for the actual circumference. After a couple rides I've concluded 2096 must be the right number even though its not the actual circ of the tire. Saturday my computer read more than 1 mile more than our ride leader in a 50 mile ride. Sunday, I noticed my computer did not mtach the cue sheet and by the end of a 56 mile ride, was, again, over a mile more.

I've reset the computer back to what the manual says (2096).

Has anyone else found this to be true and if so, I wonder why the tire size in the manual does not match the actual circ of the tire?

BTW, this is the second computer I've found this to be true - both different brands.

I'm a bit anal maybe but I want my computer to be accurate.

Three things come to mind.

The manual may have a smaller size than the calculated circumference, because when the tires have weight on them, the diameter is less. Depending on weight and air pressure.

The computer software is based on circumference, but doesn't necessarily use the exact circumference. Some early models used an inexact ratio for metric/imperial conversion for example.

The odometer function can be correct and still show greater distance than the map or another rider, if you have moved about a bit left and right and take a different line through corners.

I'd just set it to the corrected values that you came up with and call it the best that it gets.

It's trivial to do a roll-out test and get the exact number for your tire, pressure, and weight. Draw thick line across the tire with chalk. Ride up and down your driveway a couple of times. Measure the distance between the chalk marks and take the average.

I have found that computer manuals differ in what they use as circumference measurements. I have had a few computers, Cateye and Sigma, that have 2096 listed for the circumference measurement and some that show 2111. I ride on 700x25 Gatorskins and using the roll-out method with the tires inflated to 95 psi (what I always inflate them to) I get a circumference of 2105. I'm not sure how much more accurate it would be if I used the 2096 measurement vs the 2105 or if it even matters that much, but I just use 2105 and I don't worry about it.

mark your tyre with chalk .... push it one full revolution and measure .... that is the setting that you need to input into your computer

then to double check.... go for a long ride and use strava.... the distance of the ride should be very near to what strava computes, and to what your bike computer computes

Back when I cared about mileage, I'd set the circumference, ride a known distance, then correct the circumference for the difference. I don't trust either the chart or the roll-out circumference. Neither are going to be totally accurate compared to actual riding with weight on the tires.

That difference, 2135/2096, is just a 1.8% difference. (but 1.8% at 50 miles is 0.9 miles.)

I always do the rollout method. Sight downwards by the front hub to the tire valve. I try to put some weight on the bike as I roll. I use a couple of pieces of tape, aligning the valve to the first one, then sticking the second after one revolution (or even after two revolutions, and divide by 2).

The rollout test is most accurate with your full weight on the bike, so have a friend help.

Ride through something that will stick to your tire , People walking their dogs usually leave that .

measure where it comes back around and leaves a second or 3rd mark on the pavement.