The top of the print looks like fabric stretched over a frame: raised pillows between the infill lines below, often with small holes where the skin failed to close at all. The sides of the part are fine.
If the top face is complete but simply rough or ridged, that is a different fault with a different cause — see rough top surface. Pillowing has gaps and lumps; roughness does not.
The top layer is bridging, whether you meant it to or not
Below any top surface is infill, and infill is mostly air by design. The first solid layer laid over it has to span from one infill line to the next with nothing underneath in between.
If that span is short and the plastic sets quickly, the strand stays straight and the next solid layer has a flat base to build on. If the span is long or the plastic stays soft, the strand sags into the gap, its neighbours pull it sideways, and the layer above inherits a wavy foundation instead of a flat one. Each subsequent layer has less material to work with than it should, and the surface closes badly or not at all.
So a pillowed top is a bridging failure repeated a few hundred times per layer.
Do the arithmetic on solid thickness
This is where most people are simply short. A solid top needs a thickness, not a layer count, and the two only coincide at one layer height.
Aim for at least 0.8 mm of solid material on top. At 0.2 mm layers that is four layers; at 0.15 mm layers it is six; at 0.3 mm layers it is three, and those three are individually harder to lay because each is bridging further per pass. Many stock profiles ship with three top layers regardless of layer height, which is adequate at one setting and thin at another.
Most slicers now offer a minimum-thickness setting alongside the layer count, and setting it removes the problem permanently across every profile.
The support underneath
Top layers cannot be better than what they are laid on.
- Infill density. Below roughly a fifth, the gaps get wide enough that the first solid layer struggles regardless of how many follow it. Large flat tops want more than a small part does.
- Infill pattern. Patterns that produce long straight runs directly under the skin give it something continuous to land on. Gyroid, which is popular for good reasons elsewhere, presents a curved and intermittent surface to the layer above and is the weakest choice under a wide flat top.
- Ironing. Not a fix for pillowing — it works on a surface that is already closed. Run it after the problem is solved, not instead of solving it.
Cooling is the half that gets skipped
Every one of those unsupported spans has to freeze before it sags. On PLA and PETG that means the part-cooling fan at 100% for top surfaces, which is one of the few places where maximum cooling is unambiguously right.
Slowing the top solid layers helps for the same reason — a strand crossing a gap slowly has more time in contact with moving air. This is the one place in the print where slower and colder are both correct, which is worth noting because it is the opposite of the advice for layer strength.
When the material fights you
On ABS, ASA and polycarbonate the cooling advice above collides with everything those materials need to avoid cracking. There is no setting that satisfies both.
The practical answer on those materials is to attack the geometry instead: more solid layers, denser infill under the top, and a slower top surface, with cooling raised only for the final layers if the part is short enough that a brief chill will not split it. Accepting a slightly imperfect top is often the right trade on a functional ABS part.
Two habits that prevent it
Design flat tops out where you can. A top face broken up by a chamfer, a dome or a recess is far less prone than a large uninterrupted plane, because the spans are shorter and the geometry is not uniformly marginal everywhere at once.
Test on the part, not on a cube. A calibration cube's top is small and well supported. The failure shows up on the wide flat lid, which is exactly the part nobody test-prints first.
Materials that pillow
PLA and PETG are the common cases simply because they are the materials most flat-topped parts are printed in. PETG is slightly worse because its melt stays soft longer and sags further before setting. Filled and matte grades are noticeably better here, because a bead loaded with solid particles resists sagging into the gaps below it, and a wood or fibre-filled top closes where a glossy one does not.