How we measure Barcelona rents

Every figure here comes from contracts actually signed and registered with the Generalitat, not from asking prices on listing portals.

Where the numbers come from: Rents come from INCASÒL, the Catalan land institute, which publishes the average registered rent per square metre for every Barcelona district and neighbourhood each quarter. Sale prices come from the Generalitat's property transaction statistics. Both are official statistics published by the Generalitat de Catalunya.
What the average actually measures: It is the mean rent per square metre per month across contracts registered in that quarter, in that area. It reflects deals that closed, including renewals and long-standing tenancies, so it sits below current asking prices. It is the right number for checking whether a price is reasonable, and the wrong number for predicting what you will be quoted.
Why it differs from listing portals: Portals show what is being asked for the flats still on the market. Registered rents show what people agreed to pay. Asking prices in Barcelona have run close to double the registered average, so a listing well above the figures here is common. That is exactly when the official number is worth quoting.
Districts and neighbourhoods: Barcelona has 10 districts and 73 neighbourhoods. District figures are INCASÒL's own published district averages, not a mean of the neighbourhoods inside them. A plain average of a district's barris sits 2–3% off the official figure in most quarters and has been as far as 9% off since 2024, because the official one weights each barri by how many contracts it registered. Weight them that way and the two agree to within 1%. We use the official figure everywhere. The city-wide figure works the same way: it is the Barcelona total INCASÒL publishes, not an average of the ten districts, which reads about 1% lower.
Gross rental yield: Where a neighbourhood has both a rent and a sale price for the same quarter, we show gross yield: one year of rent divided by the purchase price per square metre. It is before tax, agency fees, community charges, maintenance and vacancy, so read it as a way to compare areas rather than a return you would bank.
How often it updates: INCASÒL publishes quarterly, but the exact date moves. An automated job checks the source every week, refuses to write if the spreadsheet layout has changed, and raises an alert instead of publishing something wrong. Every page shows the quarter its figures belong to.
What this data cannot tell you: It does not cover rooms in shared flats, tourist lets or seasonal contracts. Neighbourhoods with too few registered contracts in a quarter are omitted rather than estimated, so some small areas have gaps in their history. And an average says nothing about a specific flat's condition, floor or light.
Household income: Rent on its own is a price, so the barri pages also show what people there earn: the Ajuntament de Barcelona’s disposable-household-income series, published per census section under CC BY 4.0. What they do NOT show is rent divided by it. That ratio is easy to compute and hard to compute honestly: the income figure is per person while a flat houses a household of about 2.4, a fixed reference flat charges a barri where the average let is 55 m² for one of 75, and the income series ends in 2023 while rents run to the current quarter. Those errors compound in the same direction and can reverse the ranking outright. Even done properly the denominator would include owner households, understating what renters pay. So the income is published as a figure to weigh against the rent, not pre-divided into a verdict.
Corrections: If a figure here looks wrong, please tell us. The pipeline is automated and a change at the source can slip through. The full dataset behind every page is published openly, so anyone can check our arithmetic.
Letting activity counts: The barri pages show how many rental contracts INCASÒL registered in the quarter. It is a count, not a vacancy rate. Bigger barris register more contracts simply because they hold more homes (the count tracks barri size at r=0.84), so it says how much letting activity passes through an area, not how easily you will find something there. The counts come from deposits INCASÒL could place on a map, so the city total does not equal the sum of the barris. Beside it we show energy certificates issued for homes let in the barri over the last twelve months, from ICAEN: a second, faster read on the same activity.
Residents and building stock: The barri pages also carry residents, median age, licensed tourist flats and the size of the housing stock, from the Ajuntament de Barcelona under CC BY 4.0. Homes counted are the whole cadastral stock, so empty and second homes are in the denominator. None of these figures predicts rent. They describe the place you would be renting in.
What is on the street, and the four comparisons: Each barri page carries a count of ground-floor premises registered for economic activity, from the Ajuntament's commercial census (CC BY 4.0), and four further measures the city publishes: night-time road traffic noise, the share of audited slope points steeper than 6%, the share of households containing someone under 18, and the share of residents without Spanish nationality. Four things apply to all of them. The census counts premises, not quality (893 registered venues is not a verdict on whether any of them is good), and the 3,006 units that are vacant and on the market are excluded, because counting them would make emptying high streets look well supplied. Everything is a rate, never a raw count: per 1,000 residents, or as a share of its own base, since raw numbers would simply rank barris by population. The four measures have four different denominators, so they are never added together and there is no overall score; a score would have to assert that quiet is better than lively, which is the one question where that is false. And the noise figures are a propagation model driven by traffic counts and venue licences, not microphones in each barri, so they are labelled modelled wherever they appear. Where a figure rests on too few readings to publish (Can Peguera's slope sits on six audited points), it is withheld and shown as a dash, never as a zero. None of these predicts rent: 24 datasets have now been tested against price here and every one came back null.

Browse the open data