Every other number in a rental underwrite is knowable. Taxes are public record. Insurance is quotable. The mortgage is arithmetic. Rent is the one input you have to estimate, and it happens to be the one everything else depends on.
Overestimate rent by $150 per month and a "9 percent return" quietly becomes 4. Overestimate by $250 and a solid-looking duplex becomes a property you feed money every month. The margin for error on this single line is smaller than on every other line combined, which is why it deserves a real methodology instead of a glance at the listing description.
The hierarchy of rent evidence
Not all rent signals deserve equal trust. Work down this ladder and weight accordingly:
1. The actual current rent. If the property is tenant occupied, the in-place rent is ground truth for that unit today. Ask for the lease and twelve months of payment history, not just the number. A lease signed two months before listing, at a figure conveniently near the pro forma, deserves suspicion; sellers know buyers underwrite off in-place rent, and a friendly short-term lease is cheap to manufacture. A three-year tenancy with clean payments is the real thing.
One caveat cuts the other way: long-term tenants often sit below market, sometimes far below. In-place rent from a decade-long tenancy is a floor, not an estimate. That is the good direction to be wrong in, but note it, because your first turnover will reset the number and your reserves need to survive until then.
2. True comparables. Recently rented (not just listed) units within about a mile, matched on bedrooms, bathrooms, and condition. Three solid comps beat thirty loose ones.
The craft is in the matching. Adjust downward when your subject has the older kitchen, the shared laundry, the busy street, or no parking; these each move rent more than square footage does at this price point. Watch the days-on-market on your comps too: a unit that rented in four days was priced below market, and one that sat for sixty days was priced above it. And be honest about the boundary problem: in most cash-flow metros, rents change meaningfully across a few blocks, so "within a mile" can already be too generous. The right radius is the one that stays inside the same school zone and the same street character.
Listed-but-unrented "comps" are aspirations, not data. Every overpriced landlord in the ZIP code is on that list, waiting to be averaged into your spreadsheet.
3. Rent AVMs. Automated models estimate rent from comparable data at scale, and the good ones are genuinely useful, especially for screening dozens of listings where hand-comping is impractical. Two rules keep them honest. First, prefer models that return a range (low, mid, high) over a single point, because the range width is itself information: a $1,300 to $1,700 spread is the model telling you it is guessing. Second, underwrite between the low and mid values, never at the top. This is how PadSweep treats AVM data: its conservative estimate averages the model's low and mid values rather than taking the headline number, precisely because the headline number is the one that makes marginal deals look like winners.
4. HUD Fair Market Rents. HUD publishes rent data by ZIP code and bedroom count annually, and it covers essentially everywhere, including the small markets where comp tools go thin. It is slow-moving and set for a different purpose (housing assistance payment standards), and small-area FMRs tend to run high relative to what a specific ordinary unit achieves. Treat FMR as a ceiling-ish sanity check and trim it 10 to 20 percent if you must use it as an input, rather than adopting it as your estimate. The full mechanics, including the bedroom-count trap and the utilities adjustment, are in How Investors Use HUD Fair Market Rents.
5. Ratio rules of thumb. Percent-of-price shortcuts are last resorts when nothing better exists. They know nothing about your street, your finishes, or your bedroom count. If you find yourself underwriting off a ratio, what you actually have is a note to go find better data.
Whenever two rungs of the ladder disagree, the lower estimate is your underwriting number. You can be pleasantly surprised on the upside; you cannot pay the mortgage with the estimate.
Multifamily: always per unit
Duplexes and small multifamily need one extra discipline: estimate each unit separately, then add. A four-bedroom duplex is two 2-bedroom rents, not one 4-bedroom rent, and the difference is not small. In most markets, a 4-bedroom single-family rents for far less than double a 2-bedroom unit, so pricing the building by its total bedroom count overshoots badly. This is one of the most common errors in seller pro formas, and it flatters exactly the properties that need the most scrutiny.
While you are at it, confirm what the units actually are. County records and listings disagree constantly about unit counts, and an "up-down duplex" that is legally a single-family with an unpermitted second kitchen is a different purchase entirely: different rent, different insurance, different financing, different resale.
A worked example
Say you are underwriting a duplex, each unit 2 bed and 1 bath, in a working-class ZIP code:
| Source | Signal | Reading |
|---|---|---|
| In-place rent, unit A | $875, tenant of 6 years | Floor; below market |
| In-place rent, unit B | vacant | none |
| Three rented comps | $950, $975, $1,010 | Cluster near $950 to $975 |
| AVM range | $890 low, $1,020 mid, $1,150 high | Underwrite near $955 |
| ZIP-code FMR, 2 bed | $1,180 gross | Ceiling; trim before use |
The comps and the AVM's low-mid band agree around $950 per unit. FMR sitting well above both confirms there is headroom, not that headroom is your estimate. The defensible underwriting number is $950 per unit, $1,900 for the building, with unit A modeled at its actual $875 until its first turnover. If the deal only works at the FMR figure or the AVM's high value, the deal does not work.
The traps that inflate rent estimates
- The listing's own claim. "Projected rent $1,800" in a listing description is marketing by someone paid on the sale price. Verify or discard.
- Condition mismatch. Comps with renovated kitchens do not price your 1987 original. Adjust down or find better comps.
- Seasonality. Units rent higher and faster in early summer. A July comp is not a January promise, and a January vacancy lasts longer at any price.
- Concessions hiding in headline rents. "First month free" on a $1,200 listing is an $1,100 effective rent. Ask, or check how long the comp sat.
- Utilities blurring the comparison. A comp with owner-paid water against a subject where tenants pay everything is not the same rent. Normalize before comparing, especially in multifamily where owner-paid water is common.
- Section 8 confusion. Assistance programs pay up to published standards in some areas, but that does not make every unit eligible or every landlord's process identical. Underwrite market rent; treat any program premium as upside, not baseline.
The cheapest rent research nobody does
Before you write an offer on a serious candidate, spend twenty minutes acting like a tenant and like an owner.
As a tenant: search the rental listings for the ZIP code and see what your prospective tenant sees. How many competing units are on the market right now at your target rent? If there are two, you have pricing power. If there are thirty, the "market rent" from your comps is about to get negotiated downward by supply you did not model.
As an owner: call two property managers who work that neighborhood and ask what they would list each unit for, how long it would take to fill, and what concessions are currently normal. PMs quote rent numbers all day against their own leasing performance, which makes them the most grounded source available, and the call is free. If both PMs land under your comp-based estimate, believe the PMs.
After you close: protect the number
The estimate's job does not end at purchase. Re-check market rent at every lease renewal, because below-market drift compounds silently; a unit left $75 under market for four years has quietly donated thousands of dollars, and it also skews the comps for every investor after you. Move long-term tenants toward market in reasonable steps rather than all at once: a good tenant at 95 percent of market usually beats a vacancy chasing 100 percent. And keep records of what units actually achieved, because your own portfolio's rent history eventually becomes the best comp source you own.
Turn the estimate into a decision
Write your final number down next to its evidence before you run the deal, and resist revisiting it after you see what cash flow it produces. Rent estimates edited after the fact always drift in one direction, and it is not the conservative one. If the honest number kills the deal, the estimate did its job.
Once you have a defensible rent, the rest of the underwrite is mechanical: expenses, debt service, cash flow, and return on your cash. The complete sequence is in How to Analyze a Rental Property for Cash Flow, and the return math is in Cash-on-Cash Return Explained.
If you would rather not hand-build this for every listing: PadSweep layers these exact sources for every property it analyzes (actual rent when listed, AVM ranges underwritten conservatively at the low-mid average, trimmed HUD data as the wide-coverage fallback) and rejects deals whose cash flow only works at optimistic rents. Run it on your market with a free trial, or browse live underwritten deals first.