Tilting never makes it narrower
This is the fact the whole page rests on, and it is worth proving to yourself because it is so counter-intuitive. Take a cross-section of the item that measures a by b, and rotate it by an angle in the plane of the doorway. The width it now occupies is a cos θ + b sin θ. At zero degrees that is a. At ninety degrees it is b. Everywhere in between it is larger than both, because cos θ + sin θ is greater than one for every angle strictly between the two ends.
So the narrowest the item can ever present itself to a doorway is the smaller of its two cross-section dimensions, and that happens when it is square to the opening. Turning it at forty-five degrees — which is what everyone does when it will not go — makes the width worse, not better. What tilting buys you is height: if the item is too tall for the opening but comfortably narrow, rotating it trades some of that spare width for the height you need.
The 34 inch sofa and the 32 inch door
Take the case directly. A sofa 84 inches long and 34 inches deep, at a doorway with 32 inches of clear width and the usual 80 inches of height. Push it in end-first, so the face presented to the door is the 34 inch depth by the sofa's height.
- If the sofa is 32 inches tall: the cross-section is 34 by 32. Square to the door, the width is 34 and it does not fit. Rotate it ninety degrees — put the sofa on its side — and the width becomes 32, which fits exactly, and the height becomes 34, comfortably under 80. It goes through on its side. The 34 inch depth never had to fit the width, because it was free to become the height instead.
- If the sofa is 38 inches tall: the cross-section is 34 by 38. The smaller side is 34, the opening is 32, and no angle produces a width below 34. It does not go through that door in that orientation, and the arithmetic says so without ambiguity.
Now the diagonal. The diagonal of a 34 by 84 face is √(34² + 84²) = 90.6 inches, and the diagonal of the 34 by 38 cross-section is 51 inches. Neither of them is the governing number, and this is the point worth being clear about: for an item pushed straight through an opening, the item's own diagonal is not a way in. It is the largest extent of the shape, not the smallest, and rotating toward it makes the fit worse. The advice you sometimes see — "check whether the diagonal fits the door" — has the geometry backwards.
The diagonal that does govern, twice
There are two places a diagonal is genuinely the controlling number, and neither is the one people quote.
The opening's diagonal, for thin things. A doorway 32 inches by 80 has a diagonal of 86.2 inches. A flat panel — a tabletop, a mattress, a sheet of plywood, a mirror — is not confined to the plane of the door; it can be angled out of it and slid through corner to corner. That is why a 48 by 96 inch sheet of plywood goes through a normal internal door: 48 is comfortably under 86.2. The limit for a thin panel is the diagonal of the opening, and the practical requirement is floor space on both sides to swing it. This calculator applies that check only when the item's smallest dimension is six inches or less, and it treats the thickness as negligible, which is an idealisation — a six inch thick mattress genuinely does eat into the available chord.
The item's diagonal, for standing it upright. To tip a bookcase from lying down to standing up, the far corner sweeps through an arc, and the highest point of that arc is the diagonal of the face you are tipping. A bookcase 84 inches tall and 12 inches deep has a diagonal of 84.85 inches, so it needs 84.85 inches of ceiling to be stood up in place — not 84. That inch is why bookcases get assembled lying down, tipped up, and then found to be jammed against the ceiling with a corner biting into the plaster. On a stair landing with a low soffit this is frequently the constraint that decides everything.
The corner, and where the honest answer runs out
Turning a rigid rectangle through a right-angle corner between two corridors has a clean solution. For corridors of clear width a and b, and an item of width d lying flat, the longest length that can be rotated round is the minimum over all angles of a/cos θ + b/sin θ − d/(sin θ cos θ). With d at zero this reduces to the classic ladder-round-a-corner result. The consequences are harsher than people expect: two 36 inch corridors meeting at a right angle will take a 12 inch deep bookcase up to about 78 inches long, and will not turn a 34 inch deep sofa at all — the arithmetic gives a maximum length of about 34 inches, which is to say the sofa cannot be rotated in that space in any orientation lying flat.
Which is exactly why movers stand the sofa on end at the corner. Upright, the footprint stops being 84 by 34 and becomes 34 by 32, and the turn becomes trivial — provided the ceiling is high enough for the tip-up diagonal, which is the constraint that replaces it. That trade, from a length problem to a height problem, is the single most useful move in the whole business, and it is why the headroom field on this page matters more than it looks.
Beyond that the honesty has to start. This page models rigid rectangular boxes moving through rectangular openings and rotating flat through square corners. It does not model shapes that are not boxes, and it does not model tipping and turning simultaneously, which is what an experienced pair of movers actually does and which no closed formula describes. The general question — the largest shape of any form that can be manoeuvred around a corner — is the moving sofa problem, and it resisted mathematicians for more than half a century. Treat a "no" here as a strong signal to take something apart or find another route, and treat a "yes" with less than an inch of margin as a maybe, because feet, arms, castors, skirting boards and door stops all live in that inch and none of them are in the arithmetic.
Questions people ask
Will my sofa fit through a 32 inch door?
It depends on the sofa's depth and height, not its length. Sent through end-first, the face presented to the door is depth by height, and the narrowest that face can ever be is the smaller of those two numbers — rotating it does not help, because a rotated rectangle is always wider than its shorter side. So a sofa 34 inches deep and 32 inches tall goes through a 32 inch opening on its side, exactly, with the 34 inches becoming height. A sofa 34 inches deep and 38 inches tall does not go through at all, at any angle. Measure the depth and the height at the widest point including arms and feet, not the length, and remember that a 32 inch door slab gives only about 30 inches of clear gap once the stops are counted.
Does the diagonal of the furniture matter?
Not in the way it is usually stated. For an item pushed straight through a doorway, the diagonal is the largest extent of its cross-section and turning toward it makes the fit worse — the governing number is the smaller of the two cross-section dimensions. The diagonal matters in two other places. The doorway's diagonal is the limit for a thin flat panel angled through corner to corner, which is how a 4 by 8 sheet of plywood passes through an internal door. And the item's diagonal is what you need overhead to stand it up from lying down, which is why an 84 inch bookcase 12 inches deep needs 84.9 inches of ceiling rather than 84.
How much does taking the door off gain me?
Usually somewhere around an inch and three quarters, and it is free. A door standing open at ninety degrees still projects into the opening by its own thickness plus the hinge knuckle, and on a tight fit that is exactly the amount you are short by. Pull the hinge pins with a nail set and a hammer, lift the slab out, and set it aside. While the door is off, measure the true clear gap: it is between the stops on either side, not the width of the frame, and it is typically an inch to an inch and a half less than the nominal door size. If it is still tight after that, the next things to come off are the stops themselves, which are usually pinned mouldings, and then the item's own feet.
Why can movers get things round a corner that the calculator says will not fit?
Because they are not doing what the calculator models. This page treats the item as a rigid rectangle rotating flat, which is a solvable problem with a clean answer. Real movers tip the piece up on end, pivot it on one corner, tilt it back, and use the ceiling void, the stairwell above, the doorway opposite and the room beyond the corner as extra space — several of those simultaneously. That combined motion has no closed-form description, which is the honest reason the general problem is hard. Read a "no" here as: it will not go by the straightforward method, so plan on removing the legs or the arms, or on finding another route. Read a tight "yes" as a maybe, and go and put a tape on the actual doorway.