A wall that should be one thickness varies visibly along its length — thicker here, thinner there, in a pattern that has nothing to do with the model's geometry and does not repeat at any regular interval.
Only two things can cause it
The extruder converts rotation into a volume of plastic on the assumption that the filament is a cylinder of known diameter being pushed at a known rate. Both halves of that can fail.
The filament is not a constant diameter. A spool that varies along its length delivers a varying volume for the same rotation, and no setting compensates, because the slicer has no way to know.
The force is not constant. Drag that comes and goes — a binding spool, a tight bend in the path, a drive gear that grips intermittently, a partial obstruction that shifts — produces flow that comes and goes.
Distinguishing them takes calipers and ten minutes.
Measure the filament
Pull off a couple of metres and measure with calipers roughly every twenty centimetres, taking two readings at each point ninety degrees apart.
- All readings within a few hundredths of a millimetre of each other and of nominal: the filament is fine. Look at drag.
- Readings that wander by more than about 0.05 mm along the length: the diameter is your problem, and no printer adjustment will fix it.
- The two readings at each point differ from each other: the filament is oval rather than round. Oval filament rolls under the drive gear and feeds inconsistently even where the average diameter is correct. This is a common defect in cheap filament and it is invisible unless you measure in two directions.
While you are there, enter the measured average diameter in the slicer profile. Most people leave it at nominal 1.75 mm; real spools often average a little under, and that alone is a permanent few percent of error.
Then check the path
If the filament measures well, the resistance is varying. Work backwards from the nozzle:
- Does the spool turn with light finger pressure? A holder with stiff bearings, a spool rubbing on a flange, or a partially formed tangle all add drag that varies as the spool rotates.
- Does the path have a tight bend? A bowden tube routed with a sharp curve adds friction that rises and falls as the toolhead moves.
- Is the drive gear clean? Packed plastic dust in the teeth grips unevenly. A brass brush fixes it in seconds.
- Is there a partial obstruction? A fragment that moves in and out of the flow path gives exactly this signature, and it is worth a cold pull if everything upstream is clean.
What it looks like from the outside
This fault imitates several others, which is why it is worth confirming rather than assuming.
Uniform thin walls everywhere are under-extrusion, not this. Variation that repeats at a fixed vertical pitch is Z banding. Variation confined to shallow curves is a display effect rather than a flow problem — see uneven layer lines on curves.
The signature here is variation along the length of a straight run, at no regular interval, on a flat face.
Some filament simply is not good enough
This is the one page on this site whose honest first answer is sometimes to buy something else. Diameter tolerance is a manufacturing quality that is either present or not, and a spool with poor consistency will produce inconsistent walls on a perfect machine.
Manufacturers who publish a tolerance figure and hold to it are worth the difference for anything where appearance or dimensional accuracy matters. For rough functional parts, cheap filament is fine and this defect is the price of it.
Materials that are hardest to hold to size
Wood-filled and other particle-loaded grades, because the filler makes the extrusion process at the factory less uniform and the particles themselves vary. Flexible filaments, which are difficult to measure accurately with calipers at all — squeeze slightly harder and the reading changes, so measure with a light touch and take the largest consistent value.
Fibre-filled grades are usually well controlled, since they are made to a higher specification and priced accordingly. Plain PLA and PETG from established manufacturers are the most consistent thing you can buy, which is why an unexplained flow variation on those two is more likely to be drag than diameter.