How much speedometer error do 33s cause?

The 33 is only half the question

Nothing about a 33-inch tire causes error on its own. The error is the ratio between what the instrument was calibrated for and what is now bolted on, so the same 33 produces a completely different figure depending on what came off. Two real cases, both common:

• From 265/70R17 — the usual mid-size truck starting point — the reading is 4.41% low.
• From 285/70R17, already a near-33 at 32.71 in, the same tire moves the needle by 0.89%.

That is a fivefold difference between two people who both "went to 33s", which is why an answer quoted without a starting size is worth nothing.

What 4.41% looks like on the road

At an indicated 70 mph the truck is doing 73.1 mph. At an indicated 30 it is doing 31.3. The gap widens in absolute terms as speed rises, because it is a percentage rather than a fixed offset, so the error is smallest where it is least likely to matter and largest where it is most likely to be measured by someone else. The full table for this pair runs from 20 mph to 80.

The odometer moves with it, permanently

Every 100 miles the display counts, the truck has covered 104.4. Unlike the speedometer, this one accumulates: service intervals fall due later than the display suggests, and the recorded total pulls away from the real one every day the tires stay on.

Why the diameter change is +1.39 in and not two inches

A 265/70R17 stands 31.61 in, not 31, so the step up to 33.00 in is 1.39 in rather than the two inches the nicknames imply. Almost every over-estimate of speedometer error starts there — with the label rather than the measurement. The same arithmetic applied to 35s starts from a different place again.