Four. On a 0.4 mm nozzle that is four 0.45 mm lines of solid perimeter, which carries most loads without the cost curve turning against you.
The 60 mm cube at 0.2 mm layers and 20% infill:
| Walls | Plastic | Time |
|---|---|---|
| 2 | 57,824 mm³ | 3.95 h |
| 3 | 62,718 mm³ | 4.28 h |
| 4 | 67,537 mm³ | 4.6 h |
| 6 | 76,947 mm³ | 5.23 h |
Where the diminishing returns start
Each additional perimeter is a near-fixed cost — about twenty minutes and a few thousand cubic millimetres on this part — but each one contributes less than the last.
The reason is the same one that makes walls beat infill in the first place. Stress in bending falls off towards the centre of the section, so the first wall sits where stress is highest and the sixth sits well inside the part, where it is doing progressively less.
Two to four is where the money is. Four to six is a small gain for the same money again, and worth it only where you know the part is marginal.
Wall count is not wall thickness
What matters mechanically is millimetres of solid material, and that depends on the nozzle:
Four walls on a 0.4 mm nozzle lay down four 0.45 mm lines. Three walls on a 0.6 mm nozzle lay down three 0.68 mm lines — which is more solid material, in one fewer pass, on a faster print.
So a bigger nozzle reaches a given wall thickness with fewer perimeters. If you have switched nozzles and kept the wall count, you have quietly changed the part.
Design the thickness, not the count
The cleanest approach is to make wall sections an exact multiple of your line width in the model. A rib drawn at exactly four line widths slices into exactly four passes, with no gap and no thin sliver of infill trapped in the middle.
Sections that do not divide cleanly are where slicers produce gaps between perimeters and infill, and those gaps are where parts split.
Step the wall count up one at a time in the print time estimator and watch where the returns flatten on your geometry. The wall count guide works through the mechanics behind that curve.