Does infill affect print time?

Yes, more than any other single number in a slicer profile. The same 60 mm cube at 0.2 mm layers and 60 mm/s:

Zero is not free

A hollow cube still takes 1.31 hours, because the walls and the solid top and bottom skins print regardless. That is the floor, and on a small part it is most of the print.

It also explains why the curve is steepest at the start in proportional terms: going from 0% to 20% triples the time, while going from 50% to 100% not quite doubles it. The shell is a fixed cost that the first slice of infill is spread against.

Where the time actually goes

Infill is printed fast — it is straight lines with no surface-quality requirement, so slicers push it harder than perimeters. That makes it cheaper per cubic millimetre than any other feature, which is why doubling infill does not double the time even though it roughly doubles the interior volume.

The pattern matters here too. Gyroid and honeycomb involve more direction changes than a straight grid, so acceleration limits bite and the same percentage takes longer.

Where to spend the material instead

Almost nobody needs the difference between 20% and 50%. If a part is failing under load, wall count is the more efficient lever: perimeters follow the outside of the part, where bending stress is highest, while infill sits in the middle where it is lowest.

The exceptions are real but narrow — parts loaded in compression through their thickness, screw bosses, and soft filaments where infill density is effectively a hardness setting.

The interaction with layer height

Both terms multiply. A part at 0.1 mm layers and 50% infill is not a slightly slow print; it is a slow print twice over, and it is the commonest way to accidentally start a two-day job.

Step the infill field through a few values in the print time estimator and the curve above will appear for your own geometry. The money version of it is at infill and print cost, and pattern choice is covered in the infill patterns guide.