Home Projects · Startup costs
How much does it cost to build a deck?
Price your deck by the square foot, then see where it lands among what US households actually told the Census they paid. The famous $7,674 average is not a deck: it is a category that includes concrete patios, and that is why every quote you get looks high next to it.
Typical range $8,925 – $13,650
- Materials: footings, framing, boards, fasteners$3,770
- Labour, overhead and profit$4,651
- Work subcontracted out (usually the concrete)$2,079
- Total$10,500
§ 02 The return
The estimate is our cost per square foot applied to your deck. The figures beside it are what US households told the Census they actually paid to add a porch, deck, patio or terrace, in 2025 dollars. That category fuses four different structures, so treat its median as a reality check on your quote and never as the price of a deck.
Where the money goes
Recommended next steps
Some links below are affiliate links. If you buy through them, Calcatrice may earn a commission at no extra cost to you. We only suggest tools that fit your result, and a company can't pay to show up here.
You are in the range where most real decks land. Three quotes on one written scope, and ask each about the ledger connection to the house, the guardrail and the stairs, because those three are where a quote quietly grows.
By the numbers
- Census (American Housing Survey, 2023): the median household who hired a contractor to add a porch, deck, patio or terrace paid $7,674. But a quarter paid under $3,508 and the 5th percentile is $877, which tells you plainly that a large part of this category is not decks at all.
- The mean is $13,400, which is 75% above the median. That gap is the long tail of big jobs, and it is why 'average deck cost' figures vary so wildly between sites: they are quoting different points on a badly skewed distribution and calling all of them the average.
- A plain 300 sq ft pressure-treated deck, priced at our rate, comes to about $10,500 and lands at the 63rd percentile of the whole Census category. In composite it is about $16,500, at the 75th. A real deck is an above-median job here by definition, because the category is diluted with concrete patios.
- Economic Census (2022): a residential remodeller bills $102.75 per field hour in 2025 dollars and pays the person swinging the hammer $26.02. That is a measured 3.95x markup, the third-highest of every construction trade we have measured, and it rests on the lowest field wage of any of them.
- Economic Census: a remodeller's job splits 35.9% materials, 19.8% subcontracted and 44.3% labour, overhead and profit. That subcontract share is high compared with other trades, and on a deck it is usually the concrete for the footings going out to someone else.
- Census: households who built their own porch, deck or patio reported a median of $2,193, against $7,674 for hiring a contractor. Note the wording: it is the same fused category and we are not going to slip and call it a deck. Do not read it as a saving of three to one on the same job either. Decking is one of the genuinely approachable DIY builds, but the survey cannot tell you whether they built the same thing, and the structural parts (footings, ledger, guardrail) are exactly the ones that hurt people when they are wrong.
- There is no free, authoritative price per square foot for decking. The DOE measures database has none, and the federal lumber series is an index rather than a price level. Every confident per-square-foot figure you find online, including ours, is somebody's model. The difference is that ours is in a box you can change, and we show you where the result lands against what people really paid.
What is sourced, and what is ours. Nearly all of this page is sourced, and we can point at exactly the piece that is not. Sourced: what households actually paid, from the Census American Housing Survey, along with the full percentile ladder we use to place your job in that spread. Sourced and measured, not assumed: the billed rate and the way a job's dollar splits, both from the 2022 Economic Census for NAICS 236118, residential remodellers, the trade that builds most decks. They tell us a job is 35.9% materials, 19.8% subcontracted (on a deck that is usually the concrete for the footings) and 44.3% labour, overhead and profit. Ours is the cost per square foot, and it is the one that matters. Two smaller things are ours as well, and they say so where they appear: the height multiplier, and the quote-to-quote spread band. But the rate is what drives the answer. There is no free source for it, and we looked properly before saying so: we enumerated every sheet of both DOE measures workbooks, and the only RSMeans unit-cost tables in them are siding, roof insulation, duct insulation and pipe insulation. There is no decking line item anywhere. The federal lumber series is a price index, which can tell you decking got 12% dearer but never what it costs. So we put our rate in an editable box rather than burying it in the arithmetic. How much comfort should you take from where it lands? Some, but less than we first claimed. A plain pressure-treated deck at our rate lands at the 63rd percentile of the survey, which is a sensible place for a real deck to sit in a category watered down by patios and stoops. But we went and checked how sharp that test actually is, and it is blunt: our rate would have to fall below about $3 a square foot, or climb above about $205, before the placement looked wrong at all. Nearly any plausible rate passes it. So treat the placement as a sanity check that we are in the right universe, and not as evidence the rate is right to the dollar. The rate is our model. That is why it is a box you can overwrite. And a correction we owe you. This page first argued that house size and deck cost are unrelated, and pointed at two dips in the survey as proof. We were wrong, and the survey says so: bigger houses really do get dearer decks, from a median of $4,957 under 1,500 sq ft up to $13,155 above 4,000, a 2.65x spread that is statistically solid. The two dips we leaned on were noise. We had cited noise as proof of absence while a real signal sat in the same table, which is the exact trap our own rules warn about, because a refusal sounds rigorous and therefore needs a higher bar than a positive claim, not a lower one. We still do not offer a house-size selector, for the honest reason. We ask you the size of your deck, which is a measurement rather than a proxy, and multiplying a measurement by a proxy would count the same thing twice. What the house-size relationship really tracks is almost certainly your budget, not the geometry of your deck.
Sources: US Census Bureau, American Housing Survey 2023 National PUF (JOBTYPE 15, what households actually paid) · US Census Bureau, American Housing Survey (programme and documentation) · US Census Bureau, 2022 Economic Census, EC2223BASIC (the microdata behind the billed rate, NAICS 236118) · BEA, residential improvements price index (deflator)
How this estimate is calculated
- THE SURVEY CATEGORY IS FUSED, and this is the most important thing on the page. AHS JOBTYPE 15 is literally 'added porch, deck, patio, or terrace'. A poured concrete patio, a screened porch, a stoop and a cedar deck are four different jobs at four different prices, and the survey files them all in one row. Its median is therefore the middle of four blended distributions. We use it to place your job in the real spread of what people paid. We never call it the cost of a deck, and nor should anyone else.
- The cost per square foot is OUR MODEL, and it is the number that carries most of the uncertainty on this page. Two smaller assumptions are also ours and are flagged where they appear: the height multiplier and the quote-spread band. No free source publishes the rate: the DOE measures database contains no decking, and the BLS lumber series is a price index rather than a level, so it can tell you decking got dearer but never what it costs. Our rates are all-in and cover footings, framing, boards, fasteners, labour, overhead and profit. They are editable for exactly this reason.
- Where the estimate lands is a sanity check on that rate, but a weak one, and we would rather say so than dress it up. A plain 300 sq ft pressure-treated deck comes to about $10,500 and sits at the 63rd percentile of the Census category, which is a sensible place for a real deck in a distribution containing a lot of cheaper structures. But the test is blunt: the rate would have to drop below roughly $3 a square foot, or exceed roughly $205, before that placement looked wrong. Nearly any plausible rate passes. It tells you we are in the right universe. It does not tell you the rate is right to the dollar.
- The height multiplier (30% for a raised deck, 55% for a second-storey one) is our model too. Above roughly 30 inches, code requires a guardrail, the posts and footings become substantial, and you need stairs. This is the single biggest thing people forget when they price a deck from a per-square-foot figure they read somewhere.
- The split into materials, labour and subcontracting applies the 2022 Economic Census composition of a NAICS 236118 job (35.9% materials, 19.8% subcontracted, 44.3% labour, overhead and profit). Those shares are measured, but they are the average across all residential remodelling work in the country, so treat the split as indicative rather than as a costing of your deck.
- A deck may also be built by a framing contractor (NAICS 238130, which bills less per hour) or a finish carpentry contractor (238350). We use residential remodellers because they are the trade that takes on home additions and by far the largest: 128,187 establishments against 13,125 framers. If your builder is a specialist, their rate will differ.
- There is NO house-size selector on this page, and the first reason we published for that was WRONG. We claimed the relationship did not exist, citing two dips in the size-band medians. Those dips are noise (z = -0.28 and z = -0.57). The trend we overlooked is real and significant: the median runs from $4,957 for a home under 1,500 sq ft to $13,155 above 4,000, a 2.65x spread, with a rank correlation of +0.886. Bigger houses do get dearer decks. We keep the finding and drop the false argument, because citing noise as proof of absence is precisely the trap our own methodology warns about.
- We still decline the selector, for the reason that actually holds: this page asks for the size of your DECK, which is a direct measurement, while house size is only a proxy. Multiplying a measurement by a proxy would count the same thing twice, and what the house-size relationship is really capturing is almost certainly the household budget rather than the geometry of the deck. The bands are individually noisy too, with relative standard errors up to 40.8%, so even if we wanted them they would be a blunt instrument.
- The low and high band is our own estimate of quote-to-quote spread (15% below and 30% above), not a survey figure. The real spread in the survey is far wider than that, because it is measuring four different kinds of structure rather than four quotes for the same deck.