Residential rental property depreciates over 27.5 years and nonresidential real property over 39 years under MACRS. The line between them is not zoning or what the deed says. It is an income test in IRC section 168(e)(2): a building is residential rental property only if 80 percent or more of its gross rental income for the year comes from dwelling units. Everything else, offices, retail, warehouses, hotels, and most nightly-stay rentals, falls into the 39-year nonresidential class described in IRS Pub 946.
The 80 percent dwelling-unit income test
A dwelling unit is a house or apartment used to provide living accommodations. Units in a hotel, motel, or other establishment where more than half the units are used on a transient basis do not count. The test runs on gross rental income, not square footage, and it is applied year by year. If you live in any part of the property yourself, Pub 946 has you include the fair rental value of your own unit in the math.
A mixed-use example
Take a building with a depreciable basis of $1,400,000: ground-floor retail below 6 apartments. Gross rents are $160,000, with $132,000 from the apartments (82.5 percent) and $28,000 from the storefront. The building passes the 80 percent test, so the entire structure, retail space included, depreciates over 27.5 years: about $50,909 per year straight line. If the retail rent rose to $40,000 while apartment rent held at $132,000, the dwelling-unit share would drop to 76.7 percent, the building would flip to 39-year property, and the annual deduction would fall to about $35,897. That is roughly $15,000 per year of deduction riding on a few thousand dollars of rent mix.
The short-term rental trap
The most common surprise is the transient-use rule. Under Pub 946, a property where the majority of stays run under 30 days is treated like lodging, not a dwelling, so a full-time Airbnb depreciates over 39 years even though it looks like a house. Note that this is a different test from the 7-day average stay rule in Reg 1.469-1T(e)(3), which governs whether the activity is passive, not how long the building depreciates.
When the distinction matters less
Cost segregation works on either class. The 5-, 7-, and 15-year components a study carves out, carpet, cabinetry, dedicated electrical, parking, fencing, depreciate on their own schedules regardless of whether the shell is 27.5-year or 39-year property. The shell classification still matters for everything left behind, and a 39-year shell makes accelerating the short-life pieces more valuable, not less. Amadae's cost segregation screen models both the shell life and the reclassification estimate from your purchase data; it is a screening estimate, not an engineering study, but it shows which side of the line a property sits on before you spend anything.