Methodology

What goes into the score, criterion by criterion: weight, source and calculation rule. The same calculation applies to every city.

Principle

Better quality of life → higher score (0–10). Cost of living and the land price per m² are not part of the score. The scores are indicative; each metric cites its source.

Weights

Core: sun 9.94 · water 9.94 · beach/mountains 11.59 · restaurants 9.11 · bars 3.73 · parks 9.94 · entertainment 11.73
Complementary: mild climate 4.97 · air 4.14 · health 4.97 · safety 15 · airport 2.48 · culture 2.48
Safety remains 15%. On November 10, 2026 the bars weight was halved (7.45 → 3.73) and the difference was added to entertainment (8 → 11.73); the other weights did not change.

Entertainment (A1ed, 50 km radius)

OpenStreetMap via the ohsome API (HeiGIT, © OSM contributors, ODbL). Sum, with a log scale and a cap of 120→10:

Destinations: tourism=theme_park, zoo, aquarium and leisure=water_park.
Stadiums and music venues: leisure=stadium / building=stadium, amenity=music_venue, amenity=concert_hall.
Left out: events_venue, sports_centre, theatre, amusement_ride and amusement_arcade (inflation or venues of uncertain size; OSM almost never has capacity≥5,000).

Source: https://api.ohsome.org/v1 (elements/count, 50 km bcircles). The URL of each query appears on the city's profile.

Safety penalty (approved)

Violent deaths (homicides + undetermined intent) per 100,000: >10 → cap 8.0; >20 → cap 7.0; >30 → cap 6.0. Recife, Rio and Vitória are capped at 6.0.

How we measure safety

The safety score (0–10) combines three things, because homicide alone does not show robbery, theft or fear of walking in the street:

40% lethal violence: violent deaths (homicides + undetermined intent) per 100,000 inhabitants; 30 or more → 0, 0.5 or fewer → 10 (log scale).
30% robbery compared with the country itself: robberies recorded by the police in the city divided by the country average, with the same source and the same definition on both sides. Each country records crime differently, so we do not compare Rome's rate directly with Madrid's; we compare how far each city is above or below its own country. Half the average → 10; equal to the average → 7; five times the average → 0 (log scale).
30% feeling of safety: share of residents who feel safe walking alone at night. In Europe we use the European Commission's city survey (Quality of Life in European Cities 2023); where it does not exist, the 2024 Gallup survey for the whole country, with a note. 30% → 0; 95% → 10.

Without local robbery data, since November 10, 2026 a labelled estimate with the “estimated” badge is used (see Estimates). Without local violent deaths, safety stays out of the score, as before. The penalty for violent deaths has not changed. Robbery sources: Ministerio del Interior (Spain), Prefettura di Roma and Eurostat, INE/DGPJ (Portugal), ELSTAT (Greece), CBS (Netherlands), CSO (Ireland), Danmarks Statistik, SSMSI (France), Vaud Police and OFS (Switzerland), FBI CDE (United States), Statistics Canada, Ministerio del Interior (Uruguay), SIEC (Panama). Each profile shows the cut, the year and the link.

Data completed on November 10, 2026

We filled gaps with official sources whenever possible and, when those were missing, with unofficial sources (press, portals or intermediaries citing the official figure), always marked with the unofficial badge on the city's page.

Local lethal violence in Asia and Dubai: Tokyo (Metropolitan Police, 2024), Fukuoka (Ministry of Health vital statistics, 2023), Taipei (Taipei Police, 2024), Kuala Lumpur (Royal Malaysia Police via OpenDOSM, 2023), Seoul (Seoul Metropolitan Police, 2025), Bangkok and Chiang Mai (assault deaths from Thailand's Ministry of Public Health, 2021, the latest year published; for Chiang Mai, the whole province); Dubai (Dubai Police rate cited by the press) and Denpasar (Denpasar Police report cited by the press) are unofficial. In Japan, Thailand and Dubai we add undetermined-intent deaths using the WHO national ratio (a derived estimate, as we already did for Toronto and Orlando). Taiwan, Malaysia and Indonesia are not in the WHO database: only homicide is used, with a note that the definition is incomplete. In Seoul the police count includes attempted homicides, so we do not add undetermined deaths and the profile carries an incomplete-definition note.
Robbery in Brazilian cities: vehicle robbery in the city (state public-security departments) divided by the Brazilian average (SINESP/MJSP), same definition on both sides, IBGE 2024 population. Belo Horizonte, Recife and Vitória come from the press or an intermediary citing the department (unofficial) and, when the figure covers only part of the year, we compare it with Brazil over the same period. Florianópolis had no municipal figure; since the afternoon of November 10, 2026 it has a labelled estimate (see Estimates).
Robbery in Orlando: 2024 robberies from the Orlando Police, compiled by the Major Cities Chiefs Association (unofficial), divided by the US average (Bureau of Justice Statistics, 2024), Census Bureau population.
Land price per m²: Athens, Lausanne and Denpasar gained an unofficial reference value (market appraisal or a portal), with a caveat, and now also appear in the chart. Copenhagen, Toronto, Kuala Lumpur and Panama City had no comparable figure; since the afternoon of November 10, 2026 they have a labelled estimate (see Estimates).

The full record (value, unit, period, cut, link and access date of each figure, and the unsuccessful attempts) is in the project's source files.

Adjustments on November 10, 2026

Weights: bars went from 7.45 to 3.73 (half) and the difference was added to entertainment, which went from 8 to 11.73. The other weights did not change.
Restaurants and bars from two mapping sources: OpenStreetMap is built by volunteers and its coverage varies a lot from city to city (in Miami, 17 bars within 2 km of the centre in OSM, against 154 in Overture Maps). The mapping part of the score is now the average of two scores computed with the same formula as before: one from OpenStreetMap (ohsome, snapshot of July 27, 2026) and one from Overture Maps Places (release of October 21, 2026; only open places with confidence ≥ 0.5; restaurants = the restaurant category; bars = alcoholic-drink venues such as bars, pubs and breweries). Overture lists more places than OSM in almost every city, so its counts are divided by the median Overture/OSM ratio across the 39 cities (restaurants: 2.57 within 2 km and 3.41 in the municipality; bars: 2.48 and 3.20), so both sources are on the same scale. In Brazil the score is still half mapping, half CNPJ (Federal Revenue). The city is the municipality: in Miami, Miami Beach is a separate municipality and is not counted.
Land price per m² in Rome: the listing sample (caasa.it) now uses only the built-up zones (OMI bands B, C and D; there are listings only in C and D). Before, 82% of the listings were in the municipality's suburban and rural zones (band E, the agro romano), which pulled the median down. It is still an asking price in listings, unofficial, with a small sample.
Land price per m² in Miami and Orlando: FHFA is now read by ZIP code within the city limits (2022, the latest year) instead of the whole county (Miami-Dade and Orange).

New cities on November 10, 2026

Ten Latin American cities joined the World section: Santiago, Valparaíso, Buenos Aires, Córdoba, Mendoza, Medellín, Bogotá, Quito, Cuenca and Mexico City. They were the ones with the most complete data in the sources we already use (ERA5 climate, OpenStreetMap and Overture Maps, World Bank, WHO and an official local rate of violent deaths), and all of them follow the current methodology exactly.

Violent deaths: National Police + DANE (Colombia), Undersecretariat for Crime Prevention (Chile; Santiago is the Metropolitan Region, with a regional-figure note), Ministry of the Interior + INEC (Ecuador), SNIC (Argentina) and INEGI (Mexico City), all for 2024. Undetermined-intent deaths use the WHO national ratio, as in the other cities; in Argentina that ratio is abnormal and is not applied (incomplete-definition note).
Robbery compared with the country: only in Argentina (SNIC 2024, same definition in the city and the country). In the other cities, since the afternoon of November 10, 2026, a labelled estimate is used (see Estimates).
Feeling of safety: Gallup 2024, national sample, with a note.
Restaurants and bars: OpenStreetMap + Overture Maps with the same scale factor calibrated on the 39 cities on November 10, 2026, so the published scores do not change.
Land price per m²: listings (asking price, with a caveat) in Santiago, Valparaíso, Mendoza, Medellín and Bogotá. Buenos Aires, Córdoba, Quito, Cuenca and Mexico City had no comparable figure; since the afternoon of November 10, 2026 they have a labelled estimate (see Estimates).
Gaps: water (no World Bank figure for safely managed water in Argentina) and air (no WHO measurement for the whole city in Buenos Aires, Córdoba, Mendoza and Mexico City) were left out of those cities' scores; since the afternoon of November 10, 2026 they enter as a labelled estimate (see Estimates). Lima was left out: there is no official local rate of violent deaths for 2024.

European batch on November 10, 2026

Seven cities joined the World section: Bordeaux, Marseille, Lille, Rennes and Strasbourg (France), Oviedo (Spain) and Braga (Portugal). All of them follow the current methodology exactly, with data collected from scratch (ERA5 climate, OpenStreetMap and Overture Maps within the municipal boundary, terrain, World Bank, WHO, Wikidata).

Violent deaths: in France, 2025 homicides from SSMSI (Ministry of the Interior) in the department, because the commune-level base does not publish homicide, plus undetermined-intent deaths in the region (Eurostat, 2021–2023 average), with a regional-figure note. Oviedo: Ministerio del Interior 2024, for the municipality; undetermined-intent deaths use Spain's national ratio (a derived estimate, with a note). Braga: Eurostat, Norte region, as for Porto.
Robbery compared with the country: SSMSI 2025 by commune (armed robbery + violent robbery without a weapon) divided by France as a whole; Ministerio del Interior 2024 in Oviedo; INE/DGPJ 2022 in Braga.
Feeling of safety: the European survey on quality of life in cities (2023), for the city itself, in all seven.
Restaurants and bars: OpenStreetMap + Overture Maps with the same scale factor frozen since November 10, 2026.
Land price per m²: DVF (DGFiP), sales of building land in the commune in 2023–2025, with the same filter as Nice (Bordeaux and Lille have small samples, with a caveat); Oviedo: MIVAU official index for municipalities with more than 50,000 inhabitants in Asturias; Braga: Imovirtual listings (unofficial); Strasbourg is outside DVF (Alsace-Moselle) and received an estimate (see below).
Taxation: DMTO 2026 (DGFiP) and the 2025 taxe foncière voted by the commune and the intercommunal body (DGCL), now also in Nice; IBI 2025 from Spain's Ministry of Finance in Oviedo, Malaga, Madrid, Barcelona and Valencia; the Asturias ITP; in Braga, the 2026 IMI (0.32%, rate read in the press and approval in an official notice), and the national IMT and Stamp Duty.

Estimates

Since November 10, 2026, where a figure could not be collected for a city we use the best possible estimate instead of leaving a gap. An estimate is never presented as a measurement: each estimated value carries the estimated badge on the city's page, with the method, the base and the uncertainty range, and the city's score range (“range”) uses the ends of that interval. Base, formula and range of each estimate are recorded in the project's source files.

Robbery compared with the country: the median of the ratios measured in comparable cities (Brazil: Brazilian cities; Spanish-speaking Latin America: Argentina, Uruguay and Panama; Asia and Dubai: all cities measured outside Brazil, excluding city-states). Range = P25–P75 of the group.
Safely managed water: where the World Bank/JMP has no figure, basic water access in the country × the median safely-managed÷basic ratio in countries of the same World Bank region with basic access ≥ 95%. Taiwan (outside the World Bank): mean of Japan and South Korea.
Air (PM2.5): where there is no WHO station, the CAMS 2024 model × the median WHO÷CAMS ratio in comparable cities; Mexico City: station-weighted mean of the alcaldías measured by the WHO.
Land price per m² (not part of the score): the apartment price per m² outside the centre (Numbeo, unofficial) × the median land÷apartment ratio in cities of the same region with a published land price; Strasbourg: the notaries' apartment price × the DVF ratio of five French cities.

Taipei's life expectancy is not an estimate: it is the 2024 figure from Taiwan's Ministry of the Interior (via the press, marked as unofficial), because Taiwan is not in the World Bank database.

City	Variable	Method	Base	Estimate	Range	Date
Taipei	robbery (safety)	ratio from comparable cities	29 cities with measured robbery outside Brazil (all of them, no city-states)	2.13 × the country average	1.67–3.91	November 10, 2026
Fukuoka	robbery (safety)	ratio from comparable cities	29 cities with measured robbery outside Brazil (all of them, no city-states)	2.13 × the country average	1.67–3.91	November 10, 2026
Tokyo	robbery (safety)	ratio from comparable cities	29 cities with measured robbery outside Brazil (all of them, no city-states)	2.13 × the country average	1.67–3.91	November 10, 2026
Seoul	robbery (safety)	ratio from comparable cities	29 cities with measured robbery outside Brazil (all of them, no city-states)	2.13 × the country average	1.67–3.91	November 10, 2026
Chiang Mai	robbery (safety)	ratio from comparable cities	29 cities with measured robbery outside Brazil (all of them, no city-states)	2.13 × the country average	1.67–3.91	November 10, 2026
Santiago	robbery (safety)	ratio from comparable cities	5 Spanish-speaking Latin American cities with measured robbery	2.27 × the country average	2.13–3.16	November 10, 2026
Dubai	robbery (safety)	ratio from comparable cities	29 cities with measured robbery outside Brazil (all of them, no city-states)	2.13 × the country average	1.67–3.91	November 10, 2026
Valparaíso	robbery (safety)	ratio from comparable cities	5 Spanish-speaking Latin American cities with measured robbery	2.27 × the country average	2.13–3.16	November 10, 2026
Denpasar (Bali)	robbery (safety)	ratio from comparable cities	29 cities with measured robbery outside Brazil (all of them, no city-states)	2.13 × the country average	1.67–3.91	November 10, 2026
Medellín	robbery (safety)	ratio from comparable cities	5 Spanish-speaking Latin American cities with measured robbery	2.27 × the country average	2.13–3.16	November 10, 2026
Bogotá	robbery (safety)	ratio from comparable cities	5 Spanish-speaking Latin American cities with measured robbery	2.27 × the country average	2.13–3.16	November 10, 2026
Cuenca	robbery (safety)	ratio from comparable cities	5 Spanish-speaking Latin American cities with measured robbery	2.27 × the country average	2.13–3.16	November 10, 2026
Quito	robbery (safety)	ratio from comparable cities	5 Spanish-speaking Latin American cities with measured robbery	2.27 × the country average	2.13–3.16	November 10, 2026
Mexico City	robbery (safety)	ratio from comparable cities	5 Spanish-speaking Latin American cities with measured robbery	2.27 × the country average	2.13–3.16	November 10, 2026
Florianópolis	robbery (safety)	ratio from comparable cities	10 Brazilian cities with measured robbery	1.00 × the country average	0.38–1.77	November 10, 2026
Bangkok	robbery (safety)	ratio from comparable cities	29 cities with measured robbery outside Brazil (all of them, no city-states)	2.13 × the country average	1.67–3.91	November 10, 2026
Taipei	water (water stress and safe water)	comparable countries	World Bank: Japan and South Korea	55.90 % withdrawal / available resources	36.60–85.20	November 10, 2026
Chiang Mai	safely managed water	adjusted national proxy	basic access THA + 21 countries (East Asia & Pacific)	90.50 % of population	50.60–99.30	November 10, 2026
Buenos Aires	safely managed water	adjusted national proxy	basic access ARG + 17 countries (Latin America & Caribbean )	69.60 % of population	56.40–93.10	November 10, 2026
Panama City	safely managed water	adjusted national proxy	basic access PAN + 17 countries (Latin America & Caribbean )	66.90 % of population	54.30–89.60	November 10, 2026
Mendoza	safely managed water	adjusted national proxy	basic access ARG + 17 countries (Latin America & Caribbean )	69.60 % of population	56.40–93.10	November 10, 2026
Córdoba	safely managed water	adjusted national proxy	basic access ARG + 17 countries (Latin America & Caribbean )	69.60 % of population	56.40–93.10	November 10, 2026
Bangkok	safely managed water	adjusted national proxy	basic access THA + 21 countries (East Asia & Pacific)	90.50 % of population	50.60–99.30	November 10, 2026
Montevideo	safely managed water	adjusted national proxy	basic access URY + 17 countries (Latin America & Caribbean )	70.00 % of population	56.80–93.70	November 10, 2026
Braga	PM2.5 (clean air)	model calibrated on comparable cities	CAMS 2024 + WHO AAQ DB v8 (2 Portuguese cities (Lisbon and Porto))	7.88 µg/m³	6.90–8.90	November 10, 2026
Buenos Aires	PM2.5 (clean air)	model calibrated on comparable cities	CAMS 2024 + WHO AAQ DB v8 (9 Spanish-speaking Latin American cities)	11.32 µg/m³	10.20–12.50	November 10, 2026
Mendoza	PM2.5 (clean air)	model calibrated on comparable cities	CAMS 2024 + WHO AAQ DB v8 (9 Spanish-speaking Latin American cities)	9.23 µg/m³	8.30–10.20	November 10, 2026
Mexico City	PM2.5 (clean air)	aggregation of measurements in smaller areas	WHO AAQ DB v8, 5 alcaldías (2023)	20.96 µg/m³	20.10–22.40	November 10, 2026
Córdoba	PM2.5 (clean air)	model calibrated on comparable cities	CAMS 2024 + WHO AAQ DB v8 (9 Spanish-speaking Latin American cities)	8.80 µg/m³	8.00–9.70	November 10, 2026
Copenhagen	land price per m² (not in the score)	proxy (apartment × land/apartment ratio)	Numbeo (unofficial) + 12 European cities	1,574.80 US$/m²	1,093.00–2,665.00	November 10, 2026
Toronto	land price per m² (not in the score)	proxy (apartment × land/apartment ratio)	Numbeo (unofficial) + 2 US cities (FHFA)	777.83 US$/m²	438.00–1,117.00	November 10, 2026
Kuala Lumpur	land price per m² (not in the score)	proxy (apartment × land/apartment ratio)	Numbeo (unofficial) + 8 Asian cities	1,191.06 US$/m²	645.00–1,571.00	November 10, 2026
Buenos Aires	land price per m² (not in the score)	proxy (apartment × land/apartment ratio)	Numbeo (unofficial) + 6 Spanish-speaking Latin American cities	394.00 US$/m²	329.00–649.00	November 10, 2026
Córdoba	land price per m² (not in the score)	proxy (apartment × land/apartment ratio)	Numbeo (unofficial) + 6 Spanish-speaking Latin American cities	272.99 US$/m²	228.00–449.00	November 10, 2026
Quito	land price per m² (not in the score)	proxy (apartment × land/apartment ratio)	Numbeo (unofficial) + 6 Spanish-speaking Latin American cities	137.58 US$/m²	115.00–227.00	November 10, 2026
Cuenca	land price per m² (not in the score)	proxy (apartment × land/apartment ratio)	Numbeo (unofficial) + 6 Spanish-speaking Latin American cities	192.06 US$/m²	160.00–316.00	November 10, 2026
Mexico City	land price per m² (not in the score)	proxy (apartment × land/apartment ratio)	Numbeo (unofficial) + 6 Spanish-speaking Latin American cities	431.43 US$/m²	360.00–710.00	November 10, 2026
Panama City	land price per m² (not in the score)	proxy (apartment × land/apartment ratio)	Numbeo (unofficial) + 6 Spanish-speaking Latin American cities	357.10 US$/m²	298.00–588.00	November 10, 2026
Strasbourg	land price per m² (not in the score)	ratio from comparable cities in the same country	Notaires du Bas-Rhin (older apartments, € 3,380/m²) + DVF Nice/Bordeaux/Marseille/Lille/Rennes	508.26 €/m²	142.00–608.00	November 10, 2026

Regional figures (Europe)

Seven places without a municipal homicide figure use Eurostat NUTS2 (assaults + undetermined intent), with a note of regional, not municipal, figure.

Land price per m²

The chart's price axis is land (not apartments). Brazil: ITBI (São Paulo 2024 via Wayback, Belo Horizonte, Porto Alegre), Terracap (Federal District); Rio's aggregate ITBI is under review (West Zone bias) — the axis uses listings; listings with a caveat in cities without open ITBI data. Vitória (since 10 Nov 2026) is a labelled estimate: with no open ITBI data and lot listings clustered in beach neighbourhoods, the m² = mean assessed land value in the City Hall's 2025 Planta Genérica de Valores (weighted by the parcels of its 80 neighbourhoods) × the median, across neighbourhoods with listings, of the asking-price÷assessed-value ratio in the same neighbourhood. Cross-checked with FipeZap apartment prices × the land÷apartment ratio of the other state capitals. Details on the city's page.

World, collected on November 10, 2026. Where an official figure existed: DVF/DGFiP (Nice, sales of building land), URA Government Land Sales (Singapore, residential-site tenders), Dubai Land Department (residential land sales), CBS/Kadaster (Amsterdam, Noord-Holland provincial average) and the Treasury Department of Thailand (Bangkok, official appraisal). Without an official municipal source: the median of land listings (asking price) with the same cleaning as in Brazil, in Lisbon and Porto (Imovirtual), Rome (caasa.it) and Montevideo (InfoCasas). Each profile shows the source, n, the P25–P75 range, the exchange rate and the caveat. Cities without a reliable source still have no figure.

Key sources

Brazil crime: SIM/DATASUS (assaults + undetermined intent)
Feeling of safety: European Commission/Eurostat (urb_percep 2023) and Gallup Global Safety Report 2025
Brazil restaurants/bars: 50% OSM + 50% CNPJ (Receita) hybrid
Restaurants and bars: OpenStreetMap (ohsome) + Overture Maps Places (Overture Maps Foundation, CDLA Permissive 2.0)
Health in Brazil: CNES · Air: WHO AAQ DB
Amenities: OpenStreetMap

Nothing is invented. Gaps are made explicit (weight redistributed or “no data”).

Frequently asked questions

How is the Onde Viver Bem score calculated?

It is an indicative quality-of-life index from 0 to 10, built from 13 variables with published weights. Every metric cites a source. Current published weights include entertainment at 11.73%, bars at 3.73% and safety at 15%. The method was updated November 10, 2026.

Does cost of living enter the score?

No. Cost of living and the land price per m² are shown beside the score (and on the chart’s X axis for land) but are not part of it.

What does the safety penalty do?

When local data exist, safety is a composite: 40% lethal violence, 30% robbery relative to the same country and 30% feeling safe walking alone at night. A lethal-violence penalty can cap the total score (over 10 deaths per 100,000 → cap 8.0; over 20 → 7.0; over 30 → 6.0).

What is a labelled estimate?

When a figure could not be collected, the site may show a documented estimate. It is never presented as a measurement: it carries the “estimated” badge on the city page and in the technical notes. See the Estimates section.

How many cities are in the ranking?

This update (November 10, 2026) covers 56 cities, in the Brazil and World sections, in a single global table.