How the two differ in kind
Drawn in section, the two roofs are hard to tell apart, and that is not an accident. Each takes the slope of a plain gable and hinges it partway up: a steep lower segment that stands the upstairs walls close to upright, and a shallow upper segment carrying on to a ridge or a deck. The volume a gable would have wasted comes back as floor somebody can stand on, and it comes back the same way in both. If the section were the whole story this page would not need writing. The section is not the whole story, because it says nothing about how many sides of the building it applies to. A gambrel is a gable form: the break line runs dead straight from one end of the building to the other, and the two ends stay vertical walls built by the wall trade. A mansard is a hip form: the break line closes into a loop, returns at every corner, and by the time it has gone round there is no wall left on the building at all. Nearly every practical difference below is that one fact working its way outward.
Which makes the usual shorthand — mansard for the headroom — wrong on most plans, and it is worth being blunt about why. A steeper face leans in less for every metre it rises, so at any single edge a mansard does give back more floor than a gambrel does. But it meets four edges and the gambrel meets two: a gambrel's ends are full-height walls that give up nothing. Put the same break height on the same building and, on a compact or squarish plan, the barn form finishes with more of the floor usable, not less, having spent far less framing to get there. The mansard only claws that back once the building is long enough for its per-edge advantage down two long sides to outweigh the two ends it now eats — which is arithmetic rather than reputation, and the FAQ below works it out. What the mansard's extra work is reliably buying, whichever way that sum lands, is elevations. Its lower face is near enough to plumb to be built as a raked stud wall and to take windows the way a wall takes them, so the room gets light and view on all four sides and the building gains what reads from the street as another storey rather than a loft. That is a real thing to want. It is simply not headroom.
So the questions that decide this are not really questions about roofs. Do you need windows on the long elevations, or will the two ends do the job? Is there an existing wall underneath that would rather receive load close to plumb than take the outward push a leaning rafter delivers at the plate? How big is the footprint — because four hip corners cost what four hip corners cost whether they sit on a single garage or a full-length barn, and a small mansard has almost no floor to spread them over? And what is the building supposed to look like where it stands, which is partly taste and partly whoever grants permission there? Answer those four and this page is over. Leave them unanswered and no comparison of the two shapes can settle it for you, because on the drawing they are the same idea twice.
The factors that actually differ
Show
| Gambrel (barn) roof | Mansard roof | |
|---|---|---|
| How many sides break | Two. The break line runs straight from gable to gable — one plate line per side, set out once and repeated by every bent down the length of the building. Nothing about it changes between the first rafter pair and the last. | Four. The break line returns at every corner and closes on itself, so it is a loop rather than a pair of lines, and each of the four returns is a three-plane junction that has to be set out, cut and flashed in its own right. |
| What the ends of the building become | Ordinary vertical walls. Full headroom right up to the gable, the cheapest glazing on the building, and the only opening you can realistically walk a loft door through — which is why barn lofts have always put their opening in the end rather than the slope. | More roof. There is no gable and no wall, so every window and every access opening on any elevation is a dormer or a unit set into a slope. Light stops being wall area and becomes a count of items. |
| Where the floor is lost | At two edges only. The steep lower segment cuts in along the two long sides and the ends surrender nothing, which on a rectangular plan leaves most of the perimeter at full height. | At all four. The steeper face gives back a little at each edge it meets, but it is now also taking a bite out of both ends — so the per-edge win is genuine while the total goes the gambrel's way on a compact plan and the mansard's way once the building is long. |
| What the steep segment is built as | Rafters. It is a roof plane, so an opening in it is a trimmed hole: doubled trimmers at the sides, a header at the head, and a header that has quietly become a beam once the opening is wide against the bent spacing. | A raked stud wall — sole plate, studs at wall spacing, headers over every opening, a plate at the break. Openings are what stud walls are for, which is the practical reason mansard windows multiply and gambrel ones stay at the ends. |
| How the load reaches the wall below | Leaning. The lower segment is a rafter, and rafters spread; the plate line and whatever ties across it have to absorb an outward component that grows as the segment lies down. | Close to plumb. The steep face carries the break-line plate down near its own line, so what the wall below receives is much nearer to bearing than to thrust — which is what makes the form workable over walls that were never detailed to be pushed outward. |
| The covering at the top | The upper segment is normally still a steep-slope pitch, so one covering system does the whole roof and the break is a transition detail inside a single product's own instructions. | The upper is far shallower by design, often shallow enough to fall under the minimum slope its covering's manufacturer publishes, or to be a flat deck outright. That means a second low-slope system and a product-to-product junction along the break. |
| Junction count, and where water finds it | Two break lines, two eaves, two verges, one ridge — every one of them a straight run. There is no point on the roof where three planes meet at a compound angle. | Four break lines meeting at four corners, four eaves, a cornice or band all the way round, and where the top is a deck a perimeter that has to drain rather than shed. The corners are where mansards fail, and no version of the shape is without them. |
| How the cost scales | With length and area. A longer barn is more of the same bent, and the premium over a plain gable is essentially one extra plate line per side plus shorter members — a premium that thins out as the building grows. | With perimeter, and with a corner count that is four however small the plan gets. Four hips, four returns, a cornice on every side and the dormers that make the windows work are all fixed items, so the smaller the footprint the worse the mansard's premium looks against the floor it hands back. |
| Changing your mind later | A gable wall can be opened, glazed, doored or built through at any point in the building's life, because it is a wall and openings in walls are routine work for anybody. | There is no wall to open. Every later change is a trimmed opening in a steep plane, flashed like a wall intersection, and anything near a corner reopens the one detail on the roof nobody wants to touch twice. |
Which one, and when
Choose gambrel (barn) roof when…
- A rectangular plan where the two ends can carry the light and the access — the gable wall is the cheapest window on the building and the only opening a loft door fits through without becoming a dormer.
- A small footprint: a garage with a loft, a workshop, a single-bay barn. Four hip corners cost the same on a small building as on a big one, and a small building has no floor area to spread them over.
- New walls you are designing anyway, so the plate line, the loft floor and the tie across the break can be detailed together to take the lower segment's spread rather than retrofitted to it afterwards.
- One covering, one break-line plate per side, one bent repeated to the far gable — when the framing programme and the estimate both need to stay simple, this is the shape that is.
- A rural or agricultural setting, where the barn form is what the building is supposed to look like and a mansard reads as a townhouse that took a wrong turn.
Choose mansard roof when…
- You need windows on the long elevations. A near-vertical face takes them the way a wall does; the gambrel's answer to the same brief is a row of dormers, and dormers are priced and flashed one at a time.
- Adding a storey to a building that already exists, where the walls below would much rather receive load close to plumb than take an outward push at the plate they were never designed for.
- A square or near-square plan, or a room that genuinely has to reach all four edges — the case where the gambrel's two vertical ends stop being the advantage they are on a long building.
- High-wind exposure, where the gambrel's end wall is the elevation you would rather not present. Breaking the slope to win the volume also turns that end from a triangle into a five-sided panel — noticeably more standing wall, at height, on a building of the same span. The mansard has no such panel anywhere.
- An urban or period context, or a local rule the mansard's profile is the accepted answer to — a roof that reads as a roof has in various places and eras been permitted where another storey of wall was not, which is a question for your planning authority rather than for this page.
Now run your own numbers
This page holds no prices on purpose — a national average is wrong for almost every real project. Quantify both options with your dimensions and your local quotes.
Frequently asked questions
- Which one actually gives more upstairs room?
- It depends on the proportions of the plan, which is why neither shorthand survives contact with the arithmetic. Work it edge by edge. A mansard's steeper lower face leans in less for every metre of height, so at any single edge it does give back more floor. But it meets four edges and a gambrel meets two: a gambrel's ends are full-height walls that surrender nothing at all. On a square or squarish plan those two untouched ends dominate and the gambrel wins comfortably — the opposite of the usual shorthand. Stretch the same building out and the ends become a small share of the total, the mansard's per-edge advantage down the two long sides accumulates, and it overtakes; the steeper you make the mansard face, the sooner that crossover arrives, and with a very steep face it can land at quite ordinary barn proportions. So run it rather than trusting either reputation. Note that neither calculator hands you the figure you need directly — both split their answer into a lower and an upper sloped length, which is timber to order, not the horizontal ground the face consumes — so derive the two runs from your own inputs. For the gambrel the lower run is half the width times the break ratio; for the mansard it is the lower height divided by the tangent of its angle. Call them Rg and Rm. Usable floor is the full length times the width less twice Rg for the gambrel, and the width less twice Rm times the length less twice Rm for the mansard; work both and take the larger. Match the break heights before you compare, remembering the gambrel calculator has no break-height input — its break height is that lower run times the lower pitch over twelve. What the mansard's extra framing reliably buys, whichever way the floor sum lands, is windows on four sides and a load path that stacks.
- Why does the mansard calculator ask for degrees when the gambrel one asks for a pitch?
- Because a framing square stops being usable at a mansard's angles. At seventy degrees from horizontal the rise per twelve of run is around thirty-three, and stepping that out with a square over a short run compounds error faster than the run pays it back. A mansard's steep face is angle work: the sloped length is the face height divided by the sine of the angle, and the horizontal ground it consumes is the height divided by the tangent. The square transfers that angle onto the timber; it does not derive it. A gambrel's two segments both sit inside the range twelfths were invented for — a steep lower and a conventional upper — so each is ordinary rafter trigonometry over its own run, which is why that calculator asks for two pitches and a break ratio and nothing else.
- Can a mansard run out of geometry?
- Yes, and it is the one failure a gambrel cannot have. The two steep faces travel inward as they rise, each consuming its height divided by the tangent of its angle. Double that figure and set it against the building width: on a narrow building with a tall face, the two runs eat the whole width and there is nothing left for an upper slope to cross. The mansard calculator still returns a number in that case and drops its confidence flag, which is the point — being told beats finding out when forty steep members are already cut. The honest fixes are a lower break, a steeper face, or a flat deck at the top with no upper slope at all, which is a perfectly respectable mansard and means ordering a low-slope covering rather than a steep-slope one. A gambrel cannot reach that state: its break is expressed as a fraction of the half-width, so the upper segment always has the remainder to cross.
- Which one is more likely to leak?
- Count junctions instead of arguing about materials, since both roofs are covered with much the same products. A gambrel has two break lines, two eaves, two verges and one ridge, every one of them a straight run, and no point anywhere on the roof where three planes meet at a compound angle. A mansard has four break lines returning into four corners, four eaves, a cornice or band on every side, and an upper slope shallow enough that it is frequently a different covering system from the one below — which puts a product-to-product transition on the same line as the corners. That is where mansards fail, and it is not a standard of workmanship you can simply rise above, because the shape has four of those corners by definition. The compensating truth is that a near-vertical face sheds water almost as a wall does, so the steep segment itself is the least of anyone's worries on either roof.
- Can I put windows straight into the slope instead of building dormers?
- Sometimes, and what decides it is the angle rather than the shape's name. Near-vertical faces get detailed as walls: the unit sits in a framed opening with step flashing at the cheeks and a proper head detail above, which is why mansard windows are common and unremarkable. Shallower faces are roof-window territory, and roof windows are certified for a published slope range that a steep gambrel lower segment can sit above — outside the range in one direction while still being too raked to flash as a wall without thinking about it. Check the permitted slope for the unit you are actually buying before the opening is framed, because the answer changes what the opening is and not merely what goes into it. Either way the opening interrupts structure and lands on trimmers, and a wide one measured against the bent spacing has stopped being a header and become a beam somebody needs to size.
- How do I use the two calculators on the same building?
- Differently, and the difference is the two shapes again. A gambrel is one bent worked out once and repeated: enter the width, the two pitches and the break ratio, and every rafter pair down the building is that same pair of lengths multiplied by the count. A mansard takes two passes and a caveat. The steep face is identical all the way round — one height, one angle, one sloped length serving every elevation — so it is computed once and used four times. The upper slope is a hip over what is left, so the common-rafter length the calculator returns belongs to the short direction, where that upper slope actually crosses to a ridge. Neither tool cuts hips: the hip lengths at the four corners and the jacks dying into them are separate work, and on a mansard they are the single item most often missing from the order.