Land does not depreciate, so the part of your purchase price that belongs to the land comes out before a cost segregation study touches anything. The regulations say the split follows the fair market values of the land and the building at the time you bought them, and the accepted ways to show that are an appraisal, the county assessor's land-to-improvement ratio applied to your price, an arm's-length allocation in the purchase contract, or comparable land sales. The catch is that the number cuts both ways. A higher land value shrinks the study. A lower one grows it, which is why owners are tempted to push it down, and why an unsupported low land number is the fastest way to get an entire study questioned. Paving, fencing, sidewalks and landscaping count as 15-year land improvements, separate from the land. Only the dirt comes out. Get the dirt number right and everything downstream stands on it.
Tom and his wife Carol bought a 9,000 square foot medical office building on North 19th Avenue in Bozeman, Montana, in March 2025 for $1,600,000. They had read about cost segregation. They had run an online estimate that multiplied the whole $1,600,000 by a percentage and promised a first-year deduction around $450,000. Then their CPA asked a question they had not considered.
"How much of that is land?"
They had not thought about it. The whole $1,600,000 was the property, in their minds. But land does not depreciate, not one dollar of it, and before a study can do anything, the land has to be carved out. On their property the land turned out to be a quarter of the price. That quarter was gone from the depreciable side before the study began, and the online estimate had never mentioned it.
Depreciation is a deduction for property wearing out, being used up or becoming obsolete. Land does not wear out. The dirt is the same dirt in fifty years. The building on it ages and eventually needs replacing, so the building is depreciable, and so are its components. A cost segregation study works entirely on the building, its systems and the improvements to the land, never on the land itself.
The clarification that confuses people: improvements to land are not land. Parking lots, sidewalks, curbs, fencing, site lighting, retaining walls, storm drainage and landscaping all wear out, and the study examines them as 15-year land improvements. Clearing and rough grading of raw ground, by contrast, generally stay with the land because they have no determinable life. Only the ground itself and the costs inseparable from it are excluded.
So the very first thing that happens, before any component is examined, is that your purchase price gets split into two buckets. Land gets nothing. Building and improvements get everything the study works with. The size of the land bucket decides how much there is to work with.
Reg. 1.167(a)-5 sets the standard. When you buy land and a building together for one price, the price is apportioned between them in proportion to their fair market values at the time of purchase. Closing costs that are capitalized follow the same split. Four kinds of evidence meet that standard, and a defensible study uses whichever one or combination fits the property. The size of the resulting deduction has no say in the choice.
The appraisal. If you financed the purchase, an appraisal exists, and it usually breaks value into land and improvements. An independent appraiser with no stake in your deduction is strong support.
The assessor's ratio. Your county assessor splits the assessed value of every parcel into land and improvements, every year, and publishes it. Applying that ratio to your purchase price is a widely accepted method. The Tax Court has accepted it (Meiers v. Commissioner, T.C. Memo. 1982-51) and has used it to replace an owner's unsupported allocation (Nielsen v. Commissioner, T.C. Summary Opinion 2017-31). It is free and takes ten minutes to find.
The purchase contract. In negotiated commercial deals the contract sometimes allocates the price between land and building. An arm's-length allocation between a buyer and a seller with opposing interests carries weight, because the seller usually wants more on the land for recapture reasons and the buyer wants less.
Comparable land sales. A standalone analysis of what the underlying lot is worth, based on recent sales of similar land nearby, with the building's contributory value as the remainder.
None of these is binding on its own, and an examiner will look at whether they agree. When the assessor says 25 percent and the appraisal says 26, the number is settled. When the assessor says 25 and the owner says 8, the owner has a problem.
You do not need a professional to open this question, and doing it before you order anything saves a surprise.
If the two sources are within a few points of each other, you have your number and your study will document it. If they are far apart, you have a question worth answering before anyone models a deduction.
There is no standard percentage, and anyone who quotes one for every property is guessing. A rural warehouse on cheap ground may be 10 percent land. A suburban medical office like Tom's is often 20 to 30. A small retail building on a prime downtown corner can be 50 percent or more, because the lot is what the market is paying for. A mobile home park or a car wash looks like mostly land to a lender's appraiser and is mostly site improvements to an engineer, which is why those property types need the most care. The land share is a fact about your parcel and your market, and the evidence for it is already in the public record and your closing file.
The seller has a view too. On the sale side, land produces capital gain while the building carries depreciation recapture, so a seller usually wants more of the price on the land and a buyer wants less. That opposing interest is exactly why an allocation negotiated into the contract carries weight, and why a buyer who later claims a much lower land number than the one in the contract has some explaining to do.
The arithmetic is unforgiving. Depreciable basis is the purchase price, plus capitalized closing costs, less the land. Every dollar assigned to land is a dollar the study cannot reclassify. Because the land comes off the top, it shrinks every category downstream: the 5-year property, the 15-year improvements, and the 39-year building alike.
That is the lever, and it is where owners get into trouble. If a lower land value means a bigger deduction, why not assign as little as possible? Because the number has to be true. Assign 8 percent of a suburban medical office to land when the assessor and every comparable sale say 25, and you have built the whole study on a number that fails the first question an examiner asks. The paper gain evaporates and takes the credibility of every other line with it.
These numbers are illustrative and rounded. They use a 37 percent federal rate, and the actual split comes from the engineering.
First-year deduction with the study: $341,040 plus $17,829, or $358,869. Without the study, the whole $1,218,000 at 2.033 percent is $24,762. The study adds $334,107, which at 37 percent is $123,620 of federal tax. Tom qualifies as a real estate professional and materially participates in the building, so the loss is nonpassive, and he and Carol file jointly, so the Section 461(l) excess business loss cap of $626,000 for 2025, adjusted annually, sits above any loss the building produces.
The online estimate had promised about $448,000, because it multiplied the whole $1,600,000 by 28 percent. The $107,000 gap was the land. Tom and Carol were briefly disappointed by the land number and satisfied with the study.
Suppose instead the land had been pushed down to $200,000 with no support, to make the headline bigger.
Under examination, the examiner pulls the assessor's card, sees 25 percent, and resets the land to $406,000. The $57,680 is disallowed with interest. If the understatement is large enough, a 20 percent accuracy-related penalty applies. And the examiner now has a documented reason to doubt every other number in the study, including the $341,040 that was right. The defensible $341,040 you keep beats the aggressive $398,720 you lose every time, and that is the entire lesson of the land allocation.
New construction. If you bought the lot and then built, the land cost is its own closing and the question is nearly self-proving. Everything spent after the lot purchase is building or improvement, except clearing and rough grading, which generally follow the land.
Condominiums. A condo owner holds an undivided interest in the land under the project. It is small, and the study documents it from the declaration and the assessor's allocation rather than pretending it is zero.
Buying to demolish. If you buy a property intending to tear the building down, Section 280B sends the building's basis and the demolition cost into the land. There is nothing to study until the new building goes up.
An online estimate multiplies the sticker price by a percentage. It does not know the land exists. A fully engineered and accounted cost segregation study starts with the land allocation, documents its sources, and reconciles them before a single component is examined. Then it works component by component under the first two approaches the IRS Audit Techniques Guide describes: Approach 1, detailed engineering from actual cost records, and Approach 2, engineering with documented estimates reconciled to actual cost. The land number, the depreciable basis and every component below it tie to the purchase documents, so you receive a component schedule your CPA can import and the engineering support behind it, which an examiner can trace from the closing statement down. It is signed by credentialed professionals, and Cost Seg America defends it for as long as you own the building, no time limit, no hour cap, no additional fee, with written responses and participation in examiner calls alongside your CPA. You cannot win an examination on a foundation that was wrong from the first line, which is why the first line gets the most attention.
Running estimates on the full purchase price. The land is forgotten, the benefit is overstated, and the real number feels like a letdown. Start every estimate from depreciable basis.
Forcing the land number down. The single most dangerous error. An unsupported low land value invites disallowance and can sink the whole study.
Accepting a land number that is too high. A lender's appraiser working from land comparables on a property that is mostly site improvements, like a park or a car wash, can overstate the land and understate the depreciable basis. The allocation has to be supportable in both directions.
Ignoring the free evidence. The assessor's split is public. Skipping it leaves the allocation weaker than it needs to be.
Treating land improvements as land. Paving, fencing and landscaping depreciate. Leaving them in the land bucket gives away 15-year property for nothing.
A fair land allocation can take a property below the line where a study pays. If the depreciable basis after land is under $180,000, the fee is too large relative to the result, and the free proposal will say so.
A downtown lot with a small old building on it may be 60 or 70 percent land by any fair measure. The study is still valid on the building that remains, but the first-year deduction will be modest, and you should know that before you order it.
If you plan to sell within a year or two, or your losses are passive and you have no passive income to absorb them, the land allocation is the least of the reasons the study may not pencil. Those questions come first, and our guide on when a study does not pay walks through them.
Land does not depreciate, so it is removed before the study begins. The study works only on the depreciable basis that remains. A higher land value means a smaller study and a lower land value means a larger one.
By the fair market values of the land and building at purchase, shown through an appraisal, the county assessor's land-to-improvement ratio applied to your price, an arm's-length allocation in the contract, or comparable land sales. The study documents which sources were used and why.
No. The allocation must be reasonable and supportable. An artificially low land value draws scrutiny, gets reset to the assessor's ratio under examination, and puts every other number in the study in doubt.
On your county assessor's property card, which is public and free, and in the appraisal from your purchase or refinance.
No. They are land improvements, which wear out and are examined as 15-year property. Only the ground itself, and the clearing and rough grading that prepared it, stays in the land bucket.
Yes. Capitalized closing costs such as title, survey, legal and transfer tax are allocated between land and building in the same proportion as the price.
Yes. A condo owner holds an undivided interest in the land under the project. It is small, and the study documents it from the declaration and the assessor's allocation.
An arm's-length allocation between parties with opposing interests is good evidence, and in a purchase of a trade or business Section 1060 makes an agreed allocation binding on both sides. It still has to be reasonable against the appraisal and the assessor.
The study weighs them, explains the difference and documents the number it uses. A gap of a few points is normal. A gap of twenty points means one source is wrong for this property, and the reasoning has to be written down.
No. It is reality, and it protects the benefit by making the study defensible. After a fair land carve-out, a well-qualified building still produces a substantial first-year deduction, especially with 100 percent bonus depreciation for property acquired and placed in service after January 19, 2025.
Before you get attached to an online estimate, pull your county assessor's card and your appraisal and find your land share. Then send Jim Dougherty and his team the property address, the purchase price and date, and the closing statement. You will get a free proposal within 24 hours that accounts for the land allocation up front, so the number you see is the one you keep, with the method, the timeline and the fee. Studies are delivered in 30 days, with rush available. Call 1-888-365-5023 or email info@costsegamerica.com.
This article is educational and is not tax advice for your situation; confirm the treatment with your CPA before filing.
Use the calculator, see your number, and request your free, no-cost proposal - delivered in 24 hours, with your flat fee quoted upfront and no obligation.