What infill percentage should I use for a functional part?

For most functional parts, 20-25% with four or five perimeters beats 50% with two. Pick the wall count first, because in bending — which is how brackets, hooks, mounts and clips actually fail — stress is highest at the outer surface and close to zero in the middle of the part, where infill lives. Infill's real jobs are supporting the top layers, resisting crushing, and tying the walls together. Raise it above about 40% only for parts loaded in compression, parts you will drill or tap, or regions around fasteners, and raise it locally with a modifier mesh rather than across the whole model.

Is 100% infill the strongest setting?

It is the heaviest setting, and it is usually not the strongest part. Filling every void asks the extruder to put down more plastic per second than many hotends can melt at normal speeds, so you get under-extrusion, or you slow the print down and pay for it in hours. It also traps heat in the part, which increases warping on anything large, and it does nothing for the failure mode that breaks most printed parts — separation between layers. A 100% infill part still splits along the layer plane at roughly the same load as a 40% one printed in the same orientation.

Should I add walls or infill?

Walls, in almost every case where the part bends or carries a load at its surface. Two extra perimeters put plastic exactly where bending stress peaks, cost little time, and add less weight than a 20-point jump in infill density. The practical limit is geometry: a wall count that exceeds half the thickness of a thin feature just converts the feature to solid, at which point extra perimeters do nothing. Slice the same model both ways and compare the grams and the estimated hours the slicer reports — that comparison takes a minute and settles the argument for your specific part.

Which infill pattern is strongest?

For general-purpose strength per gram, gyroid and cubic are the sensible defaults because they carry load in more than one direction and do not cross themselves while printing. Grid crosses its own path at every intersection, which is where blobs and extruder pressure spikes come from on dense settings. Lines and zigzag are fast and weak in anything but compression, and concentric suits flexible materials. Pattern matters less than wall count, though — switching from grid to gyroid at the same density is a smaller change than adding two perimeters.

My part keeps snapping along a layer line. Will more infill fix it?

No. A clean break along a single horizontal plane is a layer-adhesion failure, and infill does not cross that plane any more effectively than the walls do. The fixes are orientation first — rotate the model so the load runs along the layers rather than across them — then hotter nozzle, less part cooling, and a material with better interlayer bonding. If the geometry cannot be reoriented, the part belongs in a process without a weak plane, which means an SLS or MJF nylon part from a service rather than a desktop FDM print.

When does infill actually matter?

When the load pushes into the part rather than bending it, and when something has to bite into the plastic. Spacers, feet, stands and anything that gets crushed benefit directly from density. So does a region you intend to drill, tap, press a bearing into, or fit with a threaded insert, because a drill bit entering a sparse lattice chatters and tears. The useful move is local: most slicers accept a modifier mesh that raises density only around that boss or that hole, so you pay for solid plastic in the one cubic centimetre that needs it.

Why More Infill Doesn't Make 3D Prints Stronger

The bracket snapped. You went back to the slicer, pushed infill from 20% to 60%, reprinted it overnight, and it broke in the same place under the same load — after eating two and a half times the plastic and most of a day. More infill doesn't make 3D prints stronger in the way the slider implies, because on the parts people actually print, the load almost never travels through the infill.

This post contains affiliate links. If you purchase through these links, 3D Prototyping Hub may earn a small commission at no extra cost to you.

This post also contains Amazon affiliate links. As an Amazon Associate I earn from qualifying purchases.

Find the Break Before You Touch the Slider

The shape of the failure tells you which part of the print was carrying the load. Read it first, because four of these five failures are indifferent to infill density.

What the break looks like	What was actually carrying it	What to change
Bent, then cracked from the outside surface inward	The perimeters, in bending	More walls, or a fillet at the corner
Clean split along one horizontal plane	The bond between layers	Orientation, nozzle temperature, cooling
Shattered into pieces on a drop	The material's toughness	A tougher polymer, not more of a brittle one
Crushed or flattened under a steady press	The infill, genuinely	More density, locally
Screw boss split along the threads	A few millimetres of plastic around the hole	Heat-set insert, thicker boss

Only one row of that table is an infill problem. The slider is simply the most visible control in the slicer, so it is the one that gets moved.

Where the Strength Actually Lives

Bend a beam and the stress is not spread evenly through it. It peaks at the two outer surfaces — one in tension, one in compression — and falls to effectively zero along a line through the middle. That line is where infill sits.

That single fact explains most of what follows:

Perimeters are the structure. They are a continuous, solid skin running around the outside of every layer, which is exactly where bending stress is highest. Adding two perimeters puts plastic in the loaded region.
Top and bottom layers are the other half. On a flat part loaded across its face, the solid skins act like the flanges of an I-beam and the infill acts like the web — useful, but not the thing resisting the bend.
Infill has three real jobs: holding up the top layers so they do not sag into the void, resisting crushing loads that push straight into the part, and tying the two walls together so they do not buckle independently.

None of those three scale the way the percentage implies. Doubling density mostly makes the lattice finer in a region where the stress is already low, while the same gram of plastic placed in a perimeter sits at the point of maximum stress.

The practical rule that falls out: set wall count first, infill second. Four or five perimeters at 20% infill is a stiffer, stronger, lighter and faster part than two perimeters at 50% for almost every bracket, mount, clip or hook.

The Arithmetic Your Slicer Will Do for You

You do not have to take any of that on authority, and you should not. Slice the same model three ways and read the numbers the slicer already prints:

Two perimeters, 20% infill.
Two perimeters, 60% infill.
Five perimeters, 20% infill.

Compare the estimated grams and the estimated hours. On a typical functional part, version 2 costs far more material and time than version 3 and puts almost none of it where the part broke. The numbers differ per model — that is the point of doing it on yours rather than reading a figure from somebody else's test coupon.

Then measure the printed result. Slicing five perimeters does not guarantee five perimeters landed: on a thin wall the slicer collapses what will not fit, and on a curved face a perimeter can be dropped without warning. A pass over the broken edge with digital calipers tells you the wall you really got. Where the walls came out thinner than you asked for, the related symptom and its causes are in why 3D prints have gaps between the walls.

The Weak Plane Does Not Care About Density

The most common structural failure on a desktop FDM part is a clean break along one horizontal layer line. It is a bond failure between two layers, not a failure of the plastic itself, and no infill percentage addresses it — the infill crosses that plane with exactly the same weld the perimeters do.

Orientation is the lever here, and it is worth more than every other setting combined. Rotate the model so the force runs along the layers rather than across them, even when that means supports and a worse surface finish. The full diagnosis, including when a part is simply oriented wrong on the plate, is in why 3D prints snap along layer lines, and the drop-test version of the same problem is in why 3D printed parts shatter on impact.

What To Change, In Order

Work down this list and stop when the part survives. Each step costs more than the one above it.

Reorient the part so the load runs along the layers. Free.
Add perimeters. Two more is the usual step. Cheap in time, significant in bending.
Change the geometry. A fillet at the inside corner where it keeps cracking does more than any slicer setting — sharp internal corners concentrate stress, and printed parts crack precisely there. A rib is cheaper than a solid block.
Print a wider bead. A 0.6mm nozzle lays down a thicker perimeter, so you buy wall thickness without buying print time. A 0.6mm and 0.8mm nozzle set is a few dollars; the limit is your hotend's maximum flow rate, which the manufacturer publishes per material.
Change material. Only once the four above are exhausted, because a tougher polymer printed badly is still a bad part.

On step 5, be specific. Stiffness comes from filled and engineering materials — a carbon-fibre nylon (PA-CF) or a polycarbonate is a different class of part from PLA — and both bring requirements with them. Filled filaments are abrasive and need a hardened steel nozzle, because a brass orifice erodes into a wider hole and takes your chosen extrusion width with it. Both also want a hot, still chamber rather than an open frame in a cold room: QIDI's enclosed, chamber-heated machines publish a chamber temperature and a nozzle maximum per model, and those two numbers are what to compare on — their entry model is not an enclosed high-temperature machine, so read the specification per model rather than per brand. The field is covered in our enclosed printer guide, and the filled options in the carbon fibre filament guide.

Where Infill Genuinely Earns Its Weight

Four cases, and they are worth knowing precisely because the blanket advice above does not cover them:

Compression. Spacers, feet, stands, shims, anything squashed rather than bent. Here the load does pass through the lattice, and density is the setting that answers it.
Anything you will drill, tap or ream. A drill bit entering a sparse lattice chatters, grabs and tears out the far side. Solid plastic cuts cleanly — see why 3D printed parts crack when you drill them.
Fastener bosses. The plastic around a screw is loaded in a small volume and fails there. More density helps; a heat-set threaded insert installed with a temperature-controlled soldering iron helps more, because it moves the thread into brass. The failure it prevents is covered in why 3D printed threads strip.
Large flat top surfaces. Sparse infill at a wide spacing leaves the top layers bridging too far and they dimple or tear. Raising density is one fix; more top layers is the cheaper one, and both are in why 3D print top layers have holes.

In three of those four, the part needs density in one place. Most slicers accept a modifier mesh — a box or cylinder placed over a region, with its own infill setting — so you can run 15% through the body of the part and 80% around the boss. That is the setting worth learning, far more than the global percentage.

Pattern, and the 100% Trap

Pattern is a second-order choice, but not a meaningless one. Gyroid and cubic carry load in more than one direction and never cross their own path, which keeps extruder pressure steady at high densities. Grid crosses itself at every intersection — the nozzle runs back over plastic it has already laid, which is where blobs and ticking come from on dense grid infill. Lines and zigzag are the fastest and the weakest outside pure compression. Concentric is for flexibles.

Then there is 100%. It is intuitive and it is usually wrong:

It asks the hotend for more plastic per second than many can melt, so either the extruder under-extrudes or you slow the print down and pay in hours — see why 3D prints take so long.
All that plastic keeps its heat, and a big solid part shrinks more unevenly as it cools, which means more warping and more corner lift.
It still splits along the same layer plane under the same load.

If you have reached for 100% infill, the honest reading is usually that the part is wrong for the process rather than under-filled.

When To Stop Adding Plastic

There is a point where the answer is not a setting. A part that has to carry a real load in a known direction, repeatably, in more than one or two copies, is fighting the fundamental property of desktop FDM: a weak plane between every pair of layers, whatever you fill the middle with.

Powder-bed nylon processes do not have that plane, and machined parts do not have it either. If you are on your third tuning weekend, two or three quotes cost nothing — shops are listed by process and location in the 3D printing directory, and the comparison that matters is total time to a part you trust, not the price per gram. The same arithmetic applied to parts that got larger rather than denser is in why 3D printed parts fail when you scale them up, and the creep version — a part that holds at first and sags over weeks — is in why 3D printed brackets sag.

For everything else: walls first, orientation before that, and the infill slider somewhere near the bottom of the list.

Hero photograph by Inés Álvarez Fdez on Unsplash.