Building Your Own Cabinet Boxes, the Way a Shop Would

Start at 36 inches and work down, never up from the floor

Every dimension in a base cabinet is derived from one number you do not own: the finished counter sits at 36″. Dishwashers are built to slide under it. A freestanding range’s cooking surface is set to meet it. The whole under-counter appliance world assumes it. So you do not choose 36″ — you work backwards from it.

Take a 1½″ countertop off the top and the cabinet assembly has to finish at 34½″ from the floor. Take the toe kick off the bottom — 4½″ — and what is left is a carcass 30″ tall with a door very nearly that tall on the front of it. That is the whole derivation, and it is why a base cabinet reads as a 30″ face even though the box is 34½″ of assembly.

What they search at midnight “how to build kitchen cabinets” — usually with the sheets already bought, looking for a cut list. The cut list is the last thing you need. The height stack is the first.

There are three honest ways to get the 4½″: notch it out of the two side panels so each side is an upside-down L; build the carcass as a plain 30″ rectangle and set it on a separate ladder base — a frame of 2×4 or ripped sheet stock, 4½″ tall, set back 3″ from the front; or stand each box on four adjustable levelling legs and clip a separate kick board across the front afterwards. The third is what a manufactured kitchen does, Flip’s own included: the feet take out the floor and a clip-on board hides them, so the kick is never structural. For a first kitchen, build the ladder base or buy the legs. Notched sides have to be levelled one box at a time on a floor that is not flat, and any error is locked into a panel you already cut. A ladder base gets levelled once, as a continuous platform, and every box you drop on it is level and in line by definition. It also lets the toe-kick face go on last, after the flooring, as finished material rather than a sawn edge. Shops use both; site-built kitchens almost always use the second.

You can raise a counter for a tall cook by building the base taller — a 6″ plinth gives a 37½″ counter — but the appliances do not come with you. The dishwasher hangs in a hole that is now too big, and the hob sits an inch and a half below the counter it used to be flush with. Raise the kick under the cabinets and stop it at the appliance. Wall and tall cabinets have their own fixed grid; the reference numbers live in the sizes chapter, and it is worth reading before you cut anything.

Twenty-four inches is a nominal number, and the box is not twenty-four inches

“24″ deep” describes the space the cabinet occupies from the wall forward, not the size of any part in it. Every appliance that lives in the run is built to it: a dishwasher is 24″ square in plan, a countertop is fabricated at 25½″ so that it overhangs the door face, and a range is built to a 24″ bay but deliberately deeper than it — its body runs 24 to 25″ and its door and controls stand a couple of inches proud, so the cooktop lands level with the tops either side instead of behind them. Miss it and the row of appliances sits proud of the cabinets, which is the single most visible way a kitchen announces that it was improvised.

Inside that 24″ there are three parts and they add up in this order. The side panel is 23⅞″ — that is the number Flip’s own base boxes are cut to. The missing 1/8″ is not a mistake and it is not slop: it is a deliberate gap left at the back, so a box can be pushed tight against a wall that is bowed, or has a lump of plaster in it, or has a drywall bead in the corner, without the front of the box being pushed forward. The back is the one place in a kitchen where nobody will ever see a gap.

So the front edge of the carcass lands at the nominal 24″, and then the ¾″ door goes on the front of it — it does not eat into the box — putting the door face at about 24¾″ from the wall. A 25½″ countertop then overhangs that door by about ¾″, which is the standard drip edge and the reason crumbs land on the floor instead of on the doors.

Wall cabinets do not get that treatment. A 12″ wall cabinet is a full 12″ deep, because it is not being pushed down onto a floor and pressed back against a wall that has a baseboard on it — it hangs, and any wall irregularity is dealt with by a scribe strip or an end panel at the ends of the run. Tall cabinets follow the base at 24″. The one exception worth knowing is the cabinet over a fridge, which is normally built 24″ deep rather than 12″ so that its face lines up with the fridge rather than hovering a foot in front of it.

The builder’s consequence is blunt: depth is the dimension that must be identical across every box in a run. Widths vary by design, heights are fixed, but if one carcass is 1/8″ deeper than its neighbour the door faces no longer sit in one plane, and a run of doors catching the light from a window shows a step you will never stop seeing. Cut every side panel in one session, at one saw setting, against one stop.

Widths go up in threes, even when nothing is stopping you

The first thing a self-builder wants to do is size the boxes to the wall — a 146¼″ wall becomes three cabinets of 48¾″, and the run fits perfectly with no gaps. Do not do this. Cabinet widths step in 3″ increments — 12, 15, 18, 21, 24, 27, 30, 33, 36, then 42 (nobody makes a 39) — and that grid is not a manufacturing convenience. It is the language everything else in the kitchen speaks.

A dishwasher is 24″. A range is 30″ or 36″. A fridge bay is 33″ or 36″. A sink base wants to be at least 3″ wider than the sink’s outside dimension — a 33″ sink needs a 36″ box, a 30″ sink needs a 33″ — which is why sink bases cluster at 30, 33 and 36″: those are the widths the sinks people actually buy leave room inside. Every pull-out bin, cutlery insert, tray divider and corner mechanism on the market is made to those widths. Build a kitchen of 48¾″ boxes and you have built a kitchen in a private dialect: nothing off a shelf will fit it, and in fifteen years, when a door is damaged or a drawer needs replacing, nobody — including you — will be able to buy a part for it.

The remainder goes into fillers, not into the cabinets. On that 146¼″ wall you build to 144″ — say 42 + 36 + 24 + 42 — and put the leftover 2¼″ into two filler strips, one at each end. Two, not one: the wall is not square to the run and its two ends are rarely the same. A filler is a sacrificial board you scribe to the plaster with a compass and a block plane, so two fillers give you two independent places to hide the room’s error, where one makes the far end wear all of it.

Fillers earn their place even on a wall that comes out even. In an inside corner two cabinets meeting at 90° trap each other’s doors — open one and the handle fouls the return run, or a drawer hits the adjacent door front — and 3″ of filler buys the clearance back. The same applies beside a fridge, whose door has to swing past 90° before the bins will come out.

One line of holes is what makes a home build look manufactured

This is the single decision on the page. Nearly every frameless cabinet that comes out of a factory carries the same feature: two vertical rows of 5 mm holes at 32 mm centres, drilled into the inside face of each side panel, the front row set 37 mm back from the front edge and a rear row roughly the same distance in from the back. The front row is the one with a fixed number. What pins the rear row is that it has to sit a whole number of 32 mm steps behind the front one, because a drawer runner’s fixings are spaced in multiples of 32 mm and have to land on both rows at once — put it in by eye on a custom-depth box and it is a row nothing lands on. It is called the 32 mm system, and it is the reason European cabinet hardware simply fits.

What hangs off that line is everything adjustable in the cabinet. Shelf pins drop into it. Hinge mounting plates screw to two of its holes, 32 mm apart. Side-mount drawer runners screw straight to it, and an undermount runner’s rear bracket lands on it. Wire trays and pull-out frames that fix to the cabinet side are drilled to the same grid — though the floor-mounted ones, bins and most baskets, bolt to the bottom panel instead. Drill one pattern with one jig and you have pre-fitted hardware you have not bought yet, at heights you have not chosen yet — which is what “adjustable” actually means.

Four rules make it work, and each of them is a way people get it wrong.

• Index off the front edge, every panel, every timeThe front edge is the one that is visible, the one the door registers against, and the only edge that is guaranteed straight because you cut it. Register off the back and every accumulated error in your panel depth lands on the hinge plates, where it shows as doors that do not align.The checkMark the front edge of every panel with a pencil arrow as it comes off the saw, before it gets stacked and rotated.
• The two sides of a box are mirrors, not copiesBoth side panels need the same pattern at the same heights, measured from the same end, reflected about the middle of the box. Drill them as a pair, clamped face to face with their front edges flush and their bottoms aligned, and the mirror happens by itself.The failureA shelf that rocks because one row is a hole higher than the other — and no way to fix it except a new panel.
• 32 mm is 1.26″, and a 1¼″ jig is not close enoughThe two pitches differ by about a hundredth of an inch per hole, which is nothing. Twenty holes up a base-cabinet side and the top of the row is nearly three sixteenths adrift; sixty-odd up a pantry side and it is past half an inch. That will not stop a two-screw hinge plate seating — the plate spans one interval, and a Euro screw cuts its own thread in a 5 mm hole — but it stops everything that spans several holes from landing: a drawer runner whose fixings are spaced in multiples of 32 mm, a replacement plate bought to drop into holes that already exist, a shelf drilled on the other panel with the other jig.Buy onceOne jig, metric, 32 mm, used for the whole kitchen. Mixing an imperial shelf-pin jig into the same kitchen is how a run ends up with two incompatible hole patterns.
• Use a depth stop, and drill before you assembleThe published system depth is 12 to 14 mm, which is what press-in hinge hardware expects; 9 mm is plenty if the row will only ever carry shelf pins. Either way, set the stop — a 5 mm bit that goes through a ¾″ panel has just ruined the visible outside face of a cabinet.The sequenceCut, band the front edge, drill the line, then assemble. Never the other way round — a drill in an assembled box cannot be held square.

You are allowed to skip all of this: plenty of one-off builds drill hinge plates and runners off their own jigs, put shelf pins wherever they like, and the cabinets work. What you give up is interchangeability — moving a shelf, swapping a hinge, converting a door bay to drawers, buying any accessory made in the last forty years. The line adds about ten minutes a panel, and it is the difference the eye reads as “bought”.

The back panel is a squaring device, and sometimes a structural one

Backs are where two jobs get confused. One job is squareness — a rectangle of four boards is a parallelogram waiting to happen, and something has to stop it racking. The other is hanging — on a wall cabinet, something has to transfer the weight of a loaded box into a stud. A back can do either, both, or neither, depending on how it is built.

Flip’s own boxes take the middle route, and which version you get follows the core you pick: on the particleboard box a ¼″ back sits inside ¾″ rails across the top-rear and bottom-rear, the rails about 3½″ tall; on the plywood box a thinner ⅛″ back is set into a full ¾″ plywood frame — the materials chapter has both builds panel by panel. Understand either one rather than copy it, because it tells you where the load path is. The thin back is doing dust, light and stiffness. The rail is doing the hanging. So the one thing never to do with a cabinet built this way — or with any cabinet — is drive a mounting screw through the thin panel and call it fixed. Find the rail, or fit a ledger. Chapter 04 covers hanging the finished boxes.

Whichever you choose, choose it before you cut: a groove changes the width of the top, bottom and sides, and rails change the height of the back panel. And square the box against the back rather than against your eye — a back cut dead square, dropped into the opening, is the most reliable squaring jig you will ever own, and it is a part you were cutting anyway.

A sixteenth an inch, and the place the error has to go

On a frameless cabinet the doors are the front of the kitchen. There is no face frame behind them to hide anything, so the only visible detail on the whole run is the pattern of gaps between the doors. Get the gaps even and plain boxes read as joinery. Get them uneven and beautiful doors look wrong.

The working convention is 1/16″ of reveal on every exposed edge, which makes 1/8″ between a pair of doors on the same cabinet and 1/8″ between two doors on adjacent cabinets — two sixteenths meeting. Everything else follows from that arithmetic. The glossary defines reveal, overlay and scribe if those words are new.

• 18″ cabinet, one door — 18 minus two sixteenths = a door 17⅞″ wide.
• 24″ cabinet, a pair of doors — 24 minus two sixteenths minus the 1/8″ centre gap, halved: two doors at 11⅞″.
• 30″ carcass, full-height door — 30 minus two sixteenths = 29⅞″ tall.
• Three drawer fronts in a 30″ opening — a sixteenth at the top, a sixteenth at the bottom, an eighth at each of the two joints: 3/8″ of gap in all, so the three fronts share 29⅝″, which is 9⅞″ each.

Now the part that decides whether this works: error does not average out along a run — it lands in the gaps. Suppose two of your boxes come out 1/32″ wide, an error most people would call good work, and you cut their doors from the nominal widths anyway. Each of those doors now carries an extra 1/64″ of reveal on each side, so where the two meet the gap opens to 5/32″ against the 1/8″ the rest of the run is holding. The eye is startlingly good at this. It cannot tell you a box is 1/32″ wide, but it picks out the one gap wider than the gaps beside it from across the room, and it reads a gap that tapers top to bottom instantly.

There are two legitimate answers and they belong to different builders. A shop cuts every door to the drawing, because it trusts its boxes to be right; that only works if your boxes really are within a thirty-second of each other. A first-time builder cuts each door to its own box, after that box is assembled and measured with a tape you do not lift off the wood. Slower, and it makes the doors non-interchangeable forever, but it absorbs the error where it happened rather than passing it down the run.

Either way, the run itself has to be plumb before doors go on. A cabinet that is 1/8″ out of plumb tapers every gap beside it, and no amount of hinge adjustment fixes a box that is leaning — it just moves the taper somewhere else. Level the base, plumb the boxes, then hang the doors, then adjust. European hinges give you about two millimetres of movement each way, in three directions — a tuning range, not a rescue. The hinge chapter has the three screws and what each one does.

The first cut is the one that decides the kitchen

A shop breaks sheets on a vertical panel saw or a CNC nesting router, both of which hold the sheet still and move a known straight line across it. You have neither, and the whole of DIY sheet work is about reproducing that: hold the sheet still, move a straight line. Which is exactly what a track saw does, and what a circular saw on a clamped straight edge does with two clamps and a straight offcut.

Do not try to break down full sheets on a table saw alone. A 4×8 sheet of ¾″ plywood runs 60 to 70 lb and melamine-faced particleboard nearer 90, and 32 square feet of material is 32 square feet of leverage against a fence you are holding it to with two hands. Break the sheet down where it lies — on a bed of rigid foam insulation on the floor, or on sacrificial 2×4s across sawhorses — then refine the pieces on the table saw if you have one.

Three things matter more than the tool.

• Establish one reference edge and never lose itFactory edges are usually straight, but a sheet that has been carried through a yard and stacked on a forklift often is not. Rip 1/4″ off one long edge first, mark it with an arrow, and index every subsequent cut off that edge — not off whichever edge happens to be against the fence.Why it mattersAn error in the reference edge is not an error in one part. It is the same error in every part you cut afterwards.
• Cut with a stop block, not with a tapeEight side panels must be the same length, which is a stricter requirement than being the right length. A tape measure and a pencil line give you eight lengths within about 1/32″ of each other; a stop block clamped to the fence gives you eight identical parts and does it faster.The habitCut every part of a given size across the whole kitchen before you move the fence. Each fence move is a new chance to be wrong.
• Measure your actual board, not its nameNominal ¾″ plywood measures about 23/32″, and 18 mm melamine board is 0.709″. Neither is ¾″. Design a ¾″ groove for either and the panel rattles in it; design a joint around the real thickness and it fits.Two minutesCaliper the sheet at four places before the first cut, write the number on the sheet, and use that number in every calculation.

Then there is grain, which only matters on the parts that show and matters enormously there. Plan the fronts first and let the carcass parts fall into the offcuts. On a veneered or printed sheet the grain runs along the length, so a 30″ drawer front and a 36″ door are both cut along the sheet — which limits how many fronts a sheet yields and is why they get laid out first. Keep one cabinet’s fronts together, in order, off one area of one sheet, so three drawer faces above one another read as a continuous piece of wood rather than three strangers. A shop that cares treats the whole cabinet face as a single block on the sheet.

Finally, chipping. Melamine and veneer tear out on the face the blade exits, so the finished face goes up on a track saw, where the rail’s splinter strip backs the cut line, and up on a table saw too, where the teeth come down through the top and break out underneath — which is what a zero-clearance insert is for. It is a bare circular saw, cutting upward from below, that wants the good face down. A scoring pass about 1/16″ deep fixes most of the rest, and a 60–80 tooth blade for plywood or a triple-chip grind for melamine fixes the remainder. The materials chapter has more on what the sheets are.

What actually holds a frameless box square

The choice of joint matters far less than it feels like it should. What matters is which one holds the box square while the glue sets, and which one you can execute forty times over without your accuracy drifting.

Whichever you pick, the squaring procedure is the same and it has nothing to do with a framing square. Measure both diagonals of the assembled box, corner to corner, with a tape or a pair of sticks. If they match, the box is square. If they differ, clamp across the longer diagonal and pull until they agree, then fix the back. A framing square reaches about 16″ up the side of a 30″ box. An angle wrong by a third of a degree shows there as under a tenth of an inch — the sort of gap you talk yourself out of — and still puts the far corner of the box 3/16″ out.

Two more things a shop does without thinking. Assemble the box face down on a flat floor, so the front edges of all six parts bear on one plane — that guarantees the door plane is flat even if the back is not, and the door plane is the one anybody looks at. And check for twist as well as square: a box can have matching diagonals in plan and still be wound like a propeller, which you discover when the fourth corner of the door touches nothing. Sight across the front edges, or clamp the box to a known flat surface while the glue cures.

On glue: in plywood, PVA does real structural work and the fastener is mostly a clamp. In particleboard and MDF, glue in the end grain is close to worthless, so the mechanical fastener carries the joint. That is not an argument against board — it is an argument for confirmats or dowels in board rather than a plain butt joint with screws.

The exposed edge is the tell, and the first thing to fail

Edge banding gets filed under finishing, which is why it gets rushed. On a frameless cabinet it is structural to the appearance: the front edge of every panel is the front of the cabinet. It frames every opening, it runs down both sides of every door gap, it is the leading edge of every shelf, and it is the surface your knuckles meet. A raw composite edge announces itself from across a room — a stripe of unfinished brown against a white box — and no amount of good joinery elsewhere argues with it.

It is also the first thing to fail, and there is a specific place it fails. A dishwasher lets go of a load of steam at the end of every cycle. It rises off the top of the door, condenses on the cool underside of the countertop and drifts into the gap beside the machine — which is why the manufacturer ships a strip of foil tape for the counter and says nothing about the cabinet. An unbanded particleboard edge in that position swells, goes soft, and then the face lifts. This is why a finished end panel beside a dishwasher is a real part in a cabinet order rather than a scrap of leftover sheet.

Sequencing matters more than technique. Band the front edges before you drill the 32 mm line, because the 37 mm setback is measured from the finished edge — band afterwards and every hole in the kitchen has moved by the thickness of the tape, enough to make hinge plates sit differently on adjacent boxes. And band before assembly, while the panel is flat and both ends are reachable; banding inside an assembled box is a job you attempt once.

Band every edge that will ever be seen, touched or wetted: the front edges of each carcass part, the front edge of every shelf, both edges of any panel that ends a run, and the top edge of the sides in a sink base. Then take the sharp arris off each banded corner with a fine file at 45°, working away from the panel — a square banded corner catches a sleeve, and the peel starts at the corner every time.

Doors are the hard part, and buying them is a real answer

Here is where the honest advice diverges from the satisfying advice. You can build cabinet boxes in a garage to a standard nobody will ever question. Doors are a different trade, and the gap between a home-finished door and a shop-finished one is visible from the doorway.

What a shop does that a garage cannot easily copy: it sprays doors flat, horizontal, and on both faces, in a booth with filtered air, using a catalysed finish that cures chemically rather than drying by evaporation, and then it stacks them on racks for days before anybody touches them. Each of those is doing something. Horizontal spraying is how a finish levels without running. Filtered air is why there is no dust nib in it. A catalysed finish is why a fingernail does not print it three weeks later. And curing time is why the doors survive being handled on install day.

A brushed or rollered door made in a dusty space is the most common heartbreak in a self-built kitchen, and it is not a skill problem. It is the one part of the kitchen a hand is on several times a day, in the one place the light is best, and it tends to be soft for weeks — which is exactly when it gets installed. The geometry is unforgiving too: a five-piece Shaker door is four joints and a floating centre panel, and a door 1/32″ out of square announces it against a 1/16″ reveal. A slab door is easier, being a panel with banding, but it wants an already-finished face material for the same reason.

So: build the boxes and buy the fronts. Not a compromise — it is what a great many small shops do, because a door plant presses, sprays and cures more consistently than any of them can. You give the supplier the finished sizes you worked out in section six and get back doors and drawer fronts in a factory finish. Two conditions: your boxes have to be on the standard width grid, or there is nothing standard for a supplier to size a door against; and you have to agree the reveal convention before ordering, because the door is sized from the opening minus the reveals, and a supplier working to a different convention delivers a door a sixteenth wrong on all four sides.

It helps to know what is actually made in volume. Three constructions cover nearly everything that exists as a stock product: the slab door, the five-piece Shaker with a recessed centre panel, and a frame with a glass insert. Raised panels, arched tops, beadboard and true inset are specialist work with their own lead times. Flip’s own catalogue is those same three and nothing else, which is a fair picture of what a front supplier will actually have tooling for. Drawer boxes are worth a thought too: a modern one is built around a steel drawer side that is also the runner, with panels dropped in to make the floor and the back, which is why it pulls out fully without sagging — the materials chapter shows the construction panel by panel.

And then the question of where to be fussy. Care is finite, so spend it in the two places where it cannot be recovered later:

• Repeatability over accuracy — the eight side panels being identical matters more than them being exactly 23⅞″. A run of boxes all 1/32″ small looks perfect. A run where only two of them are that small looks broken. Stop block, one session, one setting.
• Square, checked on the diagonal, box by box — before the fastener goes in, not after. It is the only error that cannot be corrected later by any means, because it is glued and screwed into the assembly.

And the short list of what genuinely cannot be fixed afterwards: the 37 mm hole setback, because every hinge plate in the kitchen references it; the width of a box whose door is already cut; grain sequence on a run of drawer fronts, because the offcuts are gone; a banded edge buried under a countertop; and a racked carcass. Everything else — a shelf in the wrong place, a door that needs adjusting, a filler that needs scribing again — is an afternoon. Get those five right and the kitchen is finished the day the doors go on.