How to Calculate ROI on a Rental Property, Step by Step

One rental building honestly produces several different return figures at once, and the people quoting them are not disagreeing about arithmetic — they are using different conventions for what counts as return and what counts as invested.

Three conventions, and all of them deserve the name

There is no single ROI formula for a rental building. There are three in common use, and each was built to answer a different question.

Capitalisation rate divides net operating income by the purchase price. It excludes financing entirely, and that exclusion is the point: it lets you compare a building someone bought with a large loan against one bought outright, because it describes the building rather than the buyer.

Cash-on-cash return divides the cash left after debt service by the cash you actually committed. It is the investor-level measure — what your money is doing this year, in this position, with this loan.

Total return starts from the cash figure and adds what never arrives as cash: the principal each payment retires, and any change in the building’s value over the period. It is the fullest picture and the easiest to flatter yourself with.

None of the three is more correct than the others. A return figure quoted without naming its convention is not a number, it is an assertion — and two people arguing about whether a building returns well are usually running different formulas rather than different facts.

What belongs above the line

The return side starts with what the building collects, not what it is supposed to collect.

Scheduled rent is the sum of the leases. Collected rent is what reached the account. The gap is vacancy, turnover and non-payment, and it is not a rounding difference — in a building with one to four units there is no scale to average it away, so a single empty month lands with full weight. Model collected rent and the vacancy allowance stops being an optimistic afterthought.

Other income belongs above the line where it genuinely arrives: laundry, parking, storage, pet fees, utility reimbursements. Then subtract the operating expenses — taxes, insurance, owner-paid utilities, maintenance, management, turnover, legal and accounting, and any registration or licensing the jurisdiction requires.

Two items are deliberately not operating expenses. Debt service is excluded from net operating income, because including it would make the figure a description of the loan rather than of the building. Capital expenditure is excluded too, because a roof is not a monthly cost — it is a purchase spread over the years it serves. Both belong in the calculation somewhere; neither belongs in the operating figure.

Two further items are commonly left off the return side that arguably belong there. The first is principal paydown: every payment retires a slice of the loan, and that slice is return even though it never appears in the bank account. The second is change in value. Both are legitimate under a total-return convention, and both are also the components most often used to make a thin cash figure read well — so name them as separate lines rather than folding them in. Depreciation is a different animal again: an accounting entry rather than cash, and a question for a CPA.

What belongs below the line

The invested side moves the answer more than any argument about the numerator does, and it gets far less scrutiny.

Under the cap rate convention, the denominator is the purchase price — or the current value, if you already hold the building. Under a cash-on-cash convention it is the cash you actually put in: the down payment, the closing costs, the immediate make-ready work, and the carrying costs before the first rent arrives. Owners routinely count only the down payment. That omission shrinks the denominator and lifts the reported return, without a word of the calculation being wrong.

A third choice matters more the longer you own the building. After several years, “invested” can mean the cash you originally committed, or the equity standing in the building today. Those produce very different figures, and the second is the honest comparison when the question is whether to keep or sell — because the equity sitting in it is capital you could deploy elsewhere, whether or not you think of it that way.

Same building, same year, three defensible denominators, three different numbers. That is why the convention has to travel with the figure.

The line items owners leave out

Four expense lines go missing often enough to be predictable, and each of them moves the answer in the same direction.

Insurance. Carried over from the seller’s bill, or estimated from a building someone once owned elsewhere. Both are the wrong number for reasons covered below.

Vacancy and turnover. Not only the rent that does not arrive, but the make-ready, the letting cost and the utilities the owner picks up while the unit sits empty. An empty period also changes what the policy does — a second cost that never appears in a spreadsheet built from lease documents.

Capital reserves. Roof coverings, the heating plant, panels, service lines and water heaters are long-lived, expensive and certain. A model with no reserve line is not a model of the building — it is a model of a good year, repeated.

Management. Whether or not you hire anyone. If you do the work yourself and book nothing for it, your figure is not comparable with one where a manager is paid — and the moment you stop, the return drops without anything about the building having changed.

Real-World Scenario: Two owners run the numbers on the same duplex in the same week. One takes the seller’s expense sheet as given — the premium that owner pays, the maintenance that owner reports, no reserve line, and no charge for the weekends the new owner would spend on it. The other rebuilds every line from something they can point at: a quote at the address on the coverage they intend to carry, the levy the county will apply after a sale rather than the one on the current bill, a reserve for the roof and the heating plant, and a management figure they would have to pay somebody else. Same building, same week, two figures far enough apart to change the decision — and only one of them survives contact with the second year.

Running it with named inputs instead of invented ones

The three formulas are short enough to write out, and writing them out is most of the work.

Cap rate is net operating income divided by the purchase price. Cash-on-cash is net operating income less annual debt service, divided by the total cash committed. Total return is that same cash figure plus the principal retired and any change in value, divided by the capital committed under whichever definition you chose.

We are deliberately not filling those in with illustrative amounts. Any figure we invented would be hypothetical, and a hypothetical on a page gets remembered as a benchmark long after the sentence labeling it is forgotten. So build the sheet with your own inputs, each traceable to something you can point at:

• Collected rent — from the actual rent roll and its history, not from the leases alone.
• Taxes — the levy the county will apply after a sale, since many jurisdictions reassess on transfer and the seller’s current bill may not survive closing.
• Insurance — a quote at the address, on the coverage you intend to carry.
• Maintenance and reserves — a per-unit and per-system allowance you could defend to somebody skeptical.
• Management — a market rate for the work, booked even if you intend to do it.
• Debt service — the rate and amortisation you can actually obtain, not a rate you read somewhere.

Then run all three conventions on the same sheet, because the disagreement between them is itself information. A building with a respectable cap rate and a poor cash-on-cash figure is telling you about the financing, not the building. The reverse tells you the opposite. And one whose total return only reads acceptably once appreciation is included is telling you its operations are not carrying it.

The one input you can make exact today

Almost every line on that sheet is an estimate until you own the building. Rents are estimates until you re-let a unit. Taxes are estimates until the assessor acts. Maintenance is an estimate until something breaks. The insurance line is the exception, and owners treat it as though it were the most estimable of the lot.

A building can be quoted at its address, on the coverage you actually intend to carry, before you make an offer — and that quote is a real number rather than a placeholder. It is also your number rather than the seller’s. A premium reflects a specific deductible, a specific set of coverage decisions, a specific claim history and the building as it was rated when that owner bought it. Two owners can insure the same duplex to meaningfully different premiums and both be correct, which is why inheriting the seller’s figure imports the seller’s decisions into your model.

The choices inside that line are yours to make and they belong in the analysis explicitly: the deductible you carry, whether the building is insured to replacement cost, and whether you carry loss of rents and for how long. That last one interacts directly with the vacancy allowance you just built — it is the coverage that answers a loss of income after a covered event, which is a different question from the vacancy you plan for. What sits behind each of those decisions is set out in landlord insurance and in what actually sets the price, and what the policy does for the structure itself is covered under property coverage.

Why this page publishes no benchmark return

We are not going to tell you what a good return looks like, and there are two separate reasons.

The first is decay. A benchmark printed here would be a snapshot of the week it was written, and nothing on this site goes back to re-check it. What counts as an acceptable return moves with borrowing costs, with what safer alternatives pay, with local tax and insurance trajectories, and with the age of the building stock in question. A published figure reads just as confidently on the day it becomes wrong as it did on the day it was right, and the reader arriving two years later gets no signal that anything moved.

The second reason does not go away with better maintenance. We are an insurance agency. We are not licensed to tell you whether a specific building is a good buy, what it is worth, what your money ought to be earning, or whether the number your sheet produced is one to act on. We can teach the arithmetic, name the inputs, and make one of those inputs exact. The deciding is your work, and your CPA’s, and your adviser’s.

Two neighboring questions sit outside this page on purpose. Whether a given building is a sound investment at all is a judgment rather than a calculation, resting on things this arithmetic does not capture. And the month-by-month mechanics of what actually lands in the account is a separate discipline from the annual ratio this page is about.

Where each input actually comes from

Every input above has a public source you can run on the day you need it, which is better than a value we could print here.

For the financing line, Freddie Mac’s Primary Mortgage Market Survey publishes an ongoing national mortgage rate series. Read it as a reference point rather than as your rate — it surveys owner-occupied lending, and a loan on a rental building generally prices differently. For a rent sanity check, HUD’s Fair Market Rent datasets give area-level figures by bedroom count. They exist to administer a federal program rather than to appraise a market, so treat them as a test of whether your assumption is plausible, never as a target rent.

For vacancy, the U.S. Census Bureau’s Housing Vacancies and Homeownership program publishes series you can read directly instead of assuming a flat allowance. For cost drift — the tendency of a maintenance or insurance line to move rather than hold still — the Bureau of Labor Statistics Consumer Price Index shows how the relevant components have actually behaved. And for the boundary between a repair you expense and an improvement you capitalise, IRS Publication 527 is the starting point and your CPA is the finishing one.

We are not reproducing current values from any of those, and that is deliberate. A number printed here is a snapshot with nothing behind it to keep it current; a source named here is a tool that gives you today’s answer.

The one line we can help you make exact costs nothing to find out. Ask us for a quote on the building you are modeling, and put a real figure in the sheet before you commit to anything resting on an estimate of it.