IKEA Kallax: inserts, dividers and legs
The Kallax is IKEA's cube shelving unit, and every one of its openings is an open square begging to hold something: a basket, a drawer, a tray, a divider. That is why it has become the most-printed piece of furniture on the planet — there is a whole ecosystem of PLA accessories that rest inside, hang off, or press-fit into that cube. But all those designs share one weak spot: they take an internal dimension on faith — one your particular Kallax may not have. A tray that fits to the millimetre on screen jams halfway in the real thing — or rattles around with two millimetres of slack. This article is about that dimension — how much to subtract from it, and why.
The cube, and the dimension that governs it
A Kallax is a grid of square melamine-faced chipboard compartments. What matters for printing is not the whole unit but a single opening: a square that is open at the front and closed by four walls — the space whatever you make has to fit into. That square mouth is the reference that governs any insert, divider or drawer. Get it right and the part goes in; get it wrong and no amount of pretty design will save it.
The inside of a Kallax cube gives about 33 x 33 cm of clear width between the walls, with a depth of roughly 38–39 cm from one face to the other. And here comes an early warning: the standard Kallax has no back panel, so the opening runs straight through — open at the front and at the back. There is no vertical rear surface for a drawer to butt against, and nothing to register the part's depth against; that dimension only limits how far the insert can go before it protrudes at the rear.
These are the numbers you will see repeated across a thousand accessories, and they are fine to start from. But treat them as a guide, not gospel: IKEA revises dimensions, wall thickness varies, and the melamine itself is not milled to the micron. Two millimetres of difference between the catalogue and your shelf is perfectly normal — and two millimetres is exactly what decides whether a part cut to the limit goes in or stays out.
| Dimension | Typical value | What you use it for |
|---|---|---|
| Clear width of the opening (width x height) | ~330 x 330 mm | Sets the outer size of the insert |
| Depth of the opening (through, no back) | ~390 mm | Limits how far in before it pokes out the back |
| Wall thickness (hollow-core board) | ~39 mm | Bearing edge for tabs and dividers |
| Front rebate of the opening | — (varies by series) | Affects stops and front trim |
From a furniture dimension to an FDM clearance
Here is the step almost everyone skips. The clear opening is 330 mm, but a 330 mm insert will not go in. It needs to leave clearance on each side to slide in without forcing, and that clearance is not optional: it is the same per-side clearance that governs any sliding fit in FDM. What changes is the scale. Here the "shaft" is a part the width of your hand, and over a wall that long, everything that pushes the dimension towards a tight fit adds up.
For an insert you push in and pull out by hand, leave on the order of 1–2 mm per side — that is, 2–4 mm less on each outer dimension than the measured opening. That sounds like a lot next to the tenths you would use on a small pin — and it is, for two physical reasons. First, the melamine walls are neither perfectly flat nor square: over 33 cm, a slight bow inward swallows your whole clearance. Second, FDM shrinkage is proportional to size; in PLA it runs about 0.2–0.3 %, so over 330 mm you are already losing close to a millimetre as it cools — and considerably more in ABS or ASA. The jump between "rubs" and "won't go in" is millimetres, not tenths, precisely because the part is so large.
There is also a complication small parts never run into: a 33 cm tray does not fit on most print beds. A 220 or 256 mm bed forces you to split the part and join it — with printed joints, dowels or screws. Each joint adds its own tolerance: split a tray into two halves and if each join drifts half a millimetre, the assembled part grows a millimetre that was not in the drawing. Spread the clearance across the joints too, not just the two outer walls, and expect the final width to be set by the assembled whole, not by each piece on its own. The method — clearance per side and its conversion to the final dimension — is the same as in Real printed clearances, only here the "tolerance stack" is the whole part.
Inserts and dividers: bearing, friction or tab
Three families of parts fit inside the cube, and each solves the fit differently. Trays and small drawers rest on the floor of the opening — the horizontal shelf, which does exist — and use the walls as a simple guide: here generous clearance works in your favour, because the floor carries the weight and the walls only keep it from wandering. Do not imagine an insert "hooking" onto a back wall: there is none — the opening runs straight through.
Racks and organisers that hang off the front edge bear on the melamine edge; there, what you measure is not the clear width but the wall thickness, which runs about 39 mm — not the 15 mm it looks by eye. And that edge is deceptive: the wall is hollow-core board, two thin sheets of melamine with a cardboard honeycomb between them, not solid timber. A tab can bear comfortably across those 39 mm, but do not count on nailing or screwing into that edge: there is no solid material inside to grip, only the honeycomb.
Vertical dividers are the interesting case, because they rest on nothing: they hold themselves against the walls. You have two ways to do it. The first is by friction: the divider is cut a hair longer than the clear width and pressed in, held by the rub of its two edges against the melamine. It works, but it is treacherous at this scale: one bow in the furniture or a little shrinkage in the part and the divider drops on its own. Melamine is hard and smooth, too, so there is little friction to be had. The second, more reliable, uses tabs or stops at the ends that bear against the inner face of the walls and spread the push; you draw the body of the divider with comfortable clearance and let the tabs, short and slightly flexible, absorb the variation in the opening. A tab that flexes a millimetre forgives a millimetre of measurement error; a rigid edge does not.
Legs: the load dictates the orientation
The legs that raise the Kallax are another story, because they stop being a fit problem and become a structural one. A leg carries the weight of the unit plus whatever you put inside it — and a full Kallax easily weighs tens of kilos per bearing point — so it is no longer enough that it fits: it has to hold without breaking or creeping over time.
The first thing is orientation on the bed. A leg works mostly in vertical compression, so print it upright, with the layers horizontal and perpendicular to the load axis. It may sound backwards — the bond between layers is FDM's weak point in tension — but in compression that bond holds well: the layers press against one another instead of pulling apart. Delamination shows up when something pulls the layers open, and that happens with bending or sideways tipping, not with well-centred vertical load.
That is why, in a well-centred leg, the enemy is not delamination but buckling — a slender column that bows out all at once — and slow crushing or creep under sustained load. Keep the weight centred over the leg's axis: the moment the load goes off to one side, bending appears — and that is where a printed leg splits along a layer line. A leg lying flat on the bed puts the layer planes parallel to the axis: it takes axial push well, but any sideways load splits it along a layer joint. At that length it also needs supports, and the surface comes out worse. If the geometry forces you to lay it down, make up for it with more perimeters and high infill in the load zone.
And give it mass. A leg is no place for thin walls or 15 % infill: raise the perimeters to four or five and the infill to 40–60 % in the column that transmits the weight, so the load is carried by solid material and not by the infill pattern. Remember, too, that PLA creeps cold under sustained load: a thin, heavily loaded PLA leg can keep yielding slowly over weeks even if it never breaks outright. Against creep, the two levers that genuinely work are geometric: plenty of cross-section and lots of perimeters. On material, do not trust PETG as a cure — it too creeps under static load and is less stiff than PLA; if you want real margin, reach for a fibre-reinforced PLA (CF/GF), or ABS/ASA for its higher glass-transition temperature, and in any case give it plenty of cross-section.
Before you print: measure, then decide the fit
The summary is short. The catalogue dimension (33 x 33 cm clear, 38–39 cm through depth) is fine for drawing the rough draft, but the number that goes into the model comes from your own tape measure on your own unit. Measure the real opening at several points, subtract the per-side clearance according to what you are making — generous for trays you pull out, tighter for dividers with a tab. And if the part does not fit on the bed, spread that clearance across the assembly joints too.
To choose how tight to make each fit — when you want it to slide, when you want it fixed — lean on Choosing the fit: clearance, transition, interference; and to turn that intent into concrete millimetres, see Real printed clearances. And if what you want is to organise bins in drawers rather than open cubes, the logic of rail retention is in IKEA Trofast: the rails that hold the tubs.