The belief that thinner layers make stronger parts is close to universal among people who have just bought a printer, and it is the wrong way round often enough to be worth taking apart carefully. Layer height is primarily a control over surface finish and time. Its effect on strength is small, it is not in the direction most people assume, and it is dwarfed by a setting almost nobody adjusts.
Three runs of the same bracket
An 80 mm bracket, 95 cm³ of plastic, three walls, 20% infill, 80 mm/s, a 0.4 mm nozzle. Only the layer height changes:
| Layer height | Layers | Estimated time |
|---|---|---|
| 0.1 mm | 800 | 11.655 h |
| 0.2 mm | 400 | 5.827 h |
| 0.3 mm | 267 | 3.915 h |
Time here is almost exactly proportional to the layer count, which is what a model that knows nothing about acceleration will always give you — a real machine loses more than this to the extra stops and starts at 0.1 mm, so treat the top row as the optimistic end. Halving the layer height in the print time estimator reproduces that near-doubling on your own geometry.
The time relationship is almost exactly inverse, which is worth internalising: halving the layer height doubles the hours, and it does so on a solid part and a hollow one alike, because it is the number of passes that changed rather than the volume of plastic. The extruded volume in all three runs is within a percent of the same figure.
That first row is the one to sit with. Eight extra hours of machine time, for a part whose job is to hold something. On a machine costed at 50 cents an hour, those eight hours are worth more than the plastic in the bracket.
What a layer interface actually is
When a fresh bead is laid on a cooling one, the two do not simply touch. The heat in the new bead raises the surface of the old one back above its glass transition, and polymer chains from each side diffuse across the boundary and entangle. The strength of that weld depends on how hot the interface got and how long it stayed hot.
That mechanism explains the whole subject. A thicker bead carries more heat into a smaller number of interfaces; a thinner bead carries less heat into a larger number of them. Cooling fans remove heat from the interface. A hotter nozzle adds it. An enclosure slows its loss.
It also explains why the layer direction is the weak one at all: within a layer the material is continuous, but between layers it is welded, and a weld is only ever as good as the diffusion that formed it.
Sorting the claims by how well they are supported
It is worth being explicit about which of these statements rest on repeated published measurement and which are workshop tradition, because they are usually recited in the same breath.
Well supported. Orientation dominates everything else. Parts loaded across the layers fail at a substantially lower stress than the same geometry loaded along them, in every material and at every layer height that has been tested. If you take one thing from this page, it is that turning the part is worth more than any setting on this list.
Reasonably supported. Thicker layers tend to equal or slightly improve interlayer strength in PLA and ABS tensile testing. The mechanism above predicts it and the measurements generally agree, though effect sizes vary between studies and between materials.
Not supported. That 0.1 mm layers make a functionally stronger part than 0.2 mm. Nothing points that way; the honest summary is that fine layers are neutral at best in the weak direction.
Genuinely contested. The magnitude, and how it interacts with cooling. Layer height, nozzle temperature, fan speed and print speed all move the same interface temperature, so studies that change one and hold the others fixed reach different numbers than studies that do not. Treat any single quoted percentage with suspicion, including one you find here.
The lever that actually works
If you want a stronger part and you are willing to change one setting, do not change layer height. Change one of these, in this order:
- Orientation. Lay the part so the load runs along the layers rather than across them. Costs nothing.
- Wall count. Perimeters carry bending load far better than infill does; wall count against infill makes that case with numbers.
- Nozzle temperature. Ten degrees hotter within the material's stored range measurably improves layer bonding. The cost is stringing and slightly softer detail.
- Cooling. Turn the fan down on materials that tolerate it. This is why ABS, printed with almost no part cooling, has better interlayer behaviour than its tensile figures suggest.
- Layer height, last, and upward rather than downward.
If a part is failing along a layer line, the diagnosis is weak layer adhesion, and a finer layer height is not on the list of fixes.
When fine layers are the right call
Thin layers earn their hours in specific places:
- Curved and sloping surfaces, where the stair-stepping is visible and sanding is impractical.
- Small parts where a 0.2 mm layer is a large fraction of a feature's height — miniatures, fine text, thin threads.
- Parts that will be painted or plated, where surface artefacts show through the finish.
- Top surfaces at shallow angles, where the layer edges read as banding.
Every one of those is an appearance requirement. That is the honest category for fine layers.
A setting policy that will serve you for years
Keep a 0.2 mm profile as your baseline and treat it as the default that every other profile is a deviation from. Move to 0.3 mm for anything functional, tall, or being printed more than once — you will lose nothing you can measure and gain a third of your evening back. Move to 0.1 or 0.15 mm only when you have looked at the part and identified a surface you care about.
And when someone tells you their bracket snapped because the layers were too thick, ask which way it was oriented on the plate. It will almost always be the answer.