Pittsburgh Short-Term Rental Census: Lawrenceville and Bloomfield

I built this census because no public, listing-level count of short-term rentals existed for these neighborhoods. InsideAirbnb, the usual open source for city-level Airbnb data, does not cover Pittsburgh, and the regional open data portal returns no short-term rental dataset.

City Council is debating short-term rental rules, and the numbers being quoted in that debate come from vendors that model their estimates. This page counts instead: one row per live listing, checked by hand, with the method and its limits stated in full.

Short-term rentals by neighborhood: the headline numbers

Every count on this page is a floor. Airbnb’s search results quietly under-deliver against their own displayed totals, and the census can only include what a search returns. The methodology section below shows how much that matters and how I bounded it.

How many Airbnb listings are in Lawrenceville and Bloomfield?

273 live listings as of August 31, 2026: 176 across the three Lawrenceville sub-neighborhoods (58 Lower, 74 Central, 44 Upper) and 97 in Bloomfield, counted against the City of Pittsburgh’s official neighborhood boundaries. By type:

Whole homes dominate at 89 percent. Purpose-listed rooms are a small slice: 21 private-room listings, 14 of them in Bloomfield.

Airbnb has no field for accessory dwelling units (an ADU is a second, smaller home on the same lot, like a converted garage or a backyard cottage), so the closest observable stand-in is the listing’s own “guest suite” property type: 5 listings, 3 in Bloomfield and 2 in Central Lawrenceville, counted inside the entire-home column.

The 8 hotel-style rooms all sit in Central Lawrenceville under a single host account; they are legitimately on the platform and inside the boundary, but they are not dwelling units, so any figure framed as housing excludes them.

Short-term rental density against the housing stock

What a 2-night Airbnb stay costs in these neighborhoods

This census does not publish a nightly rate, because the price Airbnb displays for a stay includes per-stay fees, and dividing by nights gives a number no guest ever pays. Instead it records the exact total shown for the same 2-night, 2-guest stay, on stated dates, in three observed windows:

Read together, the three windows tell one story. The midweek September window ($421 across 160 bookable listings) and the December weekend ($430 across 85) sit within $10 of each other, while the late-September weekend spikes to $1,000 with only 37 listings, 14 percent of the census, left to book.

That collapse looked like a collection error, so I re-tested it before recording it: the same search box, same method, returned 198 listings with no dates, 108 for the midweek window, and 30 for that weekend. The collapse is real booking and blocking, not a data artifact.

I report it as an observation and do not assert a cause. Each median describes the listings still bookable in its window, not the whole market, because listings booked weeks out are plausibly the more desirable ones.

Who operates the listings: host concentration

Two caveats belong next to every number in this table. A host account is just an account: whether it is one person, a company, or a manager listing for several owners is not observable, so these figures count listings per account and stop there.

And the host id is the one field a human reader can no longer verify on the listing page, so it rests on the collection method alone. The extraction method itself was human-verified on a random sample, described below.

The mid-term furnished market: 30-day minimum stays

Nightly rentals are not the whole picture. Furnished Finder lists furnished rentals with 30-day minimum stays, so the platform itself is the dividing line between short-term and mid-term: these are units held out of the ordinary rental market without being nightly rentals. In the same four neighborhoods, as of August 31, 2026:

The Airbnb and Furnished Finder counts are never added together. A single unit can be listed on both platforms, both displace their map pins, and neither publishes an address, so cross-platform de-duplication is not possible. The prices are different units too: monthly asking rent on one side, a 2-night stay total on the other.

Download the dataset

Methodology: how this census was built

How the listings were found

Airbnb search results look complete and are not, in three specific ways. First, a dated search hides listings that are already booked, so the inventory comes from any-dates searches.

Second, results stop at 270 listings per map area no matter what the count badge says, so larger areas were split into smaller boxes and each box was checked against its own stated count. Third, pagination quietly under-delivers: the widened harvest collected 435 unique listings against a displayed count of 446.

I treat the collected set, not the displayed count, as the inventory, which is why every count here is a floor.

The original search rectangles, drawn by eye, missed slivers of three of the four neighborhoods. Re-searching the full official boundary extents found 130 listings the first pass never saw, 34 of them in scope, and moved the Upper Lawrenceville count up by half again.

Eight listings from the first pass did not reappear in the wider search; rechecking them one at a time showed 7 of the 8 still live. They were coverage gaps in the search results, not turnover.

Single-search counts undercount, and the fix is to widen the boxes to the full boundary file and prove each box complete against its own count.

How each listing got its neighborhood

Airbnb displaces each listing’s map pin by roughly 150 meters for privacy. I assigned each listing to a neighborhood by point-in-polygon against the City of Pittsburgh’s official boundary file, and validated the assignment three ways. Every point fell in exactly one polygon or none. A 27-point stratified random sample was checked against OpenStreetMap’s independent boundaries, and all 27 agreed (an earlier partial check had found 2 points near the Bloomfield and Garfield line, where OpenStreetMap draws community boundaries differently; the City file is authoritative here). And because of the pin displacement, 63 of the 273 live rows carry a flag marking them close enough to a boundary that displacement could change a published figure. Neighborhood totals are reliable; any single flagged row could sit one neighborhood over.

What was verified by hand, and what was not

A fixed-seed random sample of 10 listings was checked by hand against the live pages: listing id, room type and guest capacity matched on 10 of 10, which verifies the extraction is faithful. Every row was individually rechecked for liveness on August 31, 2026.

Host figures rest on the collection method alone, since the host id is no longer readable on the listing page.

Review counts drift upward between capture and any later check; that is expected, and it is one reason every figure here is a dated snapshot from late August 2026, near the end of the summer season, not an annual average.

The market also moves under the count. Between August 25 and August 31, 2 of 275 in-scope Airbnb listings were delisted, and one Furnished Finder unit at $3,000 a month left that platform. Delisted rows are kept in the dataset and dated, never deleted, because turnover is itself a finding. On the mid-term side, widening the search box moved the median rent by only $25, from $2,325 to $2,300, a useful robustness check on the narrower first harvest.

What is excluded by design

Revenue, occupancy and booking estimates are excluded: they are not observable from listings, and estimating them requires exactly the modeling this census exists to avoid. No figure here comes from AirDNA or any vendor of modeled estimates, and Furnished Finder’s own market reports (produced with AirDNA) are not used either, only its listings.

Nights shown as unavailable are not collected, because booked here, booked elsewhere, blocked by the host, and under renovation are indistinguishable. Per-listing zoning is not published, because zoning lines follow lot lines and the median listing sits just 27 meters from one, well inside the 150-meter pin displacement.

Whether a listing was previously a long-term rental is not observable, so no such claim is made. And this is an Airbnb plus Furnished Finder census only: Vrbo and other platforms are not counted, so the short-term total is a floor for the whole market.

Listings with no reviews show a rating of zero on the platform; that is an absence, not a score, so it is recorded as blank and never averaged.

Why no city registry number appears here

Pittsburgh launched a rental registry in December 2024 (Chapter 781), but the city’s own page states that rental registry compliance remains voluntary until further notice, with no enforcement or penalties (checked on the city’s page August 31, 2026). So no enforced registration count exists to compare this census against, and the regional open data portal carries no short-term rental dataset. Whether the city holds such data internally is not knowable from the portal.

What researchers have found in other cities

For context only: these studies describe other markets, and none of their figures are blended into the counts above.

Barron, Kung and Proserpio (Marketing Science, 2021) found that a 1 percent rise in Airbnb listings raises rents about 0.018 percent and home prices about 0.026 percent in the median zip code, with the effect concentrated where owner-occupancy is lower; total housing supply was unchanged, but long-term rental supply fell.

Jin, Wagman and Zhong (NBER working paper 32537, 2024) found Chicago’s registration ordinance cut active listings 16.4 percent, but only after regulators began receiving data feeds from the platforms; that study relies on AirDNA data, which is disclosed here because this census excludes such sources for its own figures.

If you want the cost side of this same market, I priced what it takes to buy and house-hack a duplex in these exact neighborhoods in the Lawrenceville and Bloomfield cost analysis, and the wider picture is in the Pittsburgh duplex prices and rents page. All of the site’s public datasets live on the open data page.