How we calculate

Every grade on this site comes from the same calculation. No editors who ‘know’ neighbourhoods, no gut feeling — only public data, applied to all 14,667 neighbourhoods in exactly the same way.

The five scores

• Safety — crimes recorded per 1,000 people present (police, 2025): violence, burglary, total crime, property crime and vandalism. Violence covers everything CBS files under it, including threats and homicide; property crime covers theft from businesses, sheds and cars, pickpocketing and shoplifting as well — everything except residential burglary, which has its own measure alongside and would otherwise count twice. The denominator counts residents plus business establishments (×3), because shops, offices and nightlife attract visitors that plain per-resident figures ignore. All crime sits in this score and nowhere else.
• Amenities — distance to supermarket, GP, train station, restaurant, hospital, café, library and motorway ramp. CBS measures those as the crow flies, so without the detour around a railway line or a river; supermarket, GP and station weigh heaviest, the motorway ramp the least.
• Wealth & income — average property value (WOZ, the heaviest measure), welfare density, the share of high and low incomes, vacancy (the more empty homes, the lower the grade) and the make-up of the housing market: owner-occupied and housing-association homes weigh the same, in opposite directions. It describes the neighbourhood, not the people living in it.
• Quiet & nature — five measures, each carrying a fifth of this score: population density, noise level, green surroundings, trees nearby and share of water. Noise, greenery and trees come from RIVM, which maps them nationwide per hectare; CBS does not publish them per neighbourhood. Greenery and trees are deliberately two measures: the green map counts all vegetation, farmland and pasture included, while the tree map counts tree canopy only — a village beside a meadow and a leafy avenue are not the same thing. Vacancy sat here until 14 August 2026 and now counts towards Wealth & income; until 19 August 2026 this score was called ‘Quiet & space’.
• Families — distance to primary school (the heaviest measure), the share of households with children, distance to childcare and secondary school, and the share of children under 15. No crime measure: violence and residential burglary sat here until 19 August 2026, but now that Families counts towards the total, crime would reach the same grade through two scores. This score measures what is there for children; how safe it is sits one score up.

From position to grade

Every grade on this site is a place in a queue, not a sum. For each measure we work out where a neighbourhood stands relative to every other neighbourhood in the Netherlands. We weigh those positions inside a score, and then once more for the total: the total score weighs the five scores as follows — Safety 20% · Amenities 20% · Wealth & income 20% · Quiet & nature 20% · Families 20% — and the grade you see is the place the neighbourhood takes with that mix among all 13,075 graded neighbourhoods. That place becomes a Dutch school grade between 2.5 and 10: the median neighbourhood gets about a 6.2, the top a 10, the bottom a 2.5.

This site mentions a slightly different neighbourhood count in different places, and that is not a contradiction — it is three counts of the same set, each serving a different purpose. CBS counts 14,667 neighbourhoods in total; every one of them has its own page on this site, including a business estate or harbour without a grade. Of those, 13,379 are residential — the gate most rankings use, because an empty office park does not belong among ‘the best neighbourhoods’. And across all neighbourhoods combined, 13,075 have enough CBS data for a total grade; that is the denominator behind every ‘no. X of Y’ on this site. A ranking or a national rank is therefore always about that last, smallest group — never about all 14,667.

So the total score is not the mean of the five grades above it — add them up and you will almost never land on the same number. That is a choice, not sloppiness. Average the five and a 9.3 for amenities cancels out a 3.3 for quiet, right across the country: 78% of all neighbourhoods would end up somewhere between 5.0 and 7.0. Everything clumps around the middle and the grade stops meaning anything. Using the national position instead, 28% falls in that band and the grade keeps answering what it promises: how many neighbourhoods in the Netherlands do better than this one, on this mix?

Because positions are evenly spread, many neighbourhoods share the same grade — 76 possible grades across 13,075 neighbourhoods, about 174 each. That is why every ranking shows the national rank next to the grade, and sorts on the underlying position rather than the rounded grade.

Shift the weighting yourself under Your mix and we redo exactly the same calculation with your weights — including that last step. Your grade therefore sits on the same scale as the official one beside it, not on a scale of its own.

What ‘Top 6% in the Netherlands’ means

Under every total grade sits one line giving the national position in plain words. For a neighbourhood in the better half you read ‘Top 6% in the Netherlands’: of all 13,075 graded neighbourhoods, fewer than six in a hundred do better. For one in the weaker half the same line counts from the other end — ‘Bottom 30% in the Netherlands’ — because nobody reads ‘top 71%’ correctly. One kind of statement, two directions.

We always round up. A neighbourhood at place 400 of 13,075 sits at 3.06% and so gets ‘top 4%’. Ordinary rounding would claim ‘top 3%’, which is not true; the same holds at the bottom. The exact place is on the page as well, in the ‘no. X of Y neighbourhoods with a grade’ line beside the name.

A district or municipality says something different: ‘Its neighbourhoods average the top 22% nationally’. A district has no position of its own among other districts — its grade is the population-weighted mean of the positions of the neighbourhoods inside it. So the line reads: the average resident of this district lives in a neighbourhood that ranks in the top 22% of the country.

What each measure contributes

First per theme: what one score is made of. The percentages below are the share within that score and add up to 100% per block.

And then the whole model in one table: score weight × measure weight, added up.

Every measure has one fixed direction, and it is in the name wherever you meet it as a choice — in Buurtwijzer and when you expand a score on the comparison page you read ‘Few welfare recipients’, ‘Many owner-occupied homes’, ‘Supermarket nearby’. Where a number follows we use the plain name (‘Welfare recipients: 24.3 per 1,000’), because ‘Few welfare recipients: 24.3’ would contradict itself; there the direction is in the colour of the bar. Name and calculation direction come from the same table in the dataset, so they cannot drift apart.

Two things to know about this weighting

Welfare and social housing track each other closely (r = 0.81). Both are in, together worth 4.2% of the overall grade. They are not the same thing — receiving benefits differs from living in a housing-association home — and the housing market belongs in from both sides: owner-occupancy up, social housing down, at exactly the same weight. Still, anyone who spots the link deserves to hear it from us.

Population density is partly the mirror image of Amenities. Density correlates −0.70 with distance to a supermarket: we reward proximity in one score and charge for the density that makes it possible in another. Now that vacancy has moved to Wealth & income, density carries more than three quarters of Space & quiet and weighs heavier than before. Within a score called ‘Space & quiet’ the direction is right, and the two largely cancel each other out in the total score: both weigh 20%. It remains a choice, not a law of nature, which is why it is stated here.

What we do not measure

• The CBS distances to green space — no values at all. Distance to park, forest, public green, open nature and sports grounds: zero of 14,667neighbourhoods carry one. Greenery and tree canopy have counted towards the grade since 19 August 2026, but through RIVM’s nationwide maps (within 300 metres of the neighbourhood), not through those CBS distances.
• Income per resident (only 13% of neighbourhoods), employment rate, the share around the social minimum and education level: zero or too few values.
• Distance to daily shops: zero values.
• Address density (near-identical to population density), the share of occupied homes (the mirror image of vacancy), the share of new-build (says little about wealth) and a few amenities that near-duplicate others (pharmacy, cinema, swimming pool, after-school care): removed to avoid double counting.

Does a neighbourhood show ‘no grade’? Then CBS publishes too little for it: we require at least 35% of the weight inside a score to be present, and at least three of the five scores for a total. A missing measure never pushes a grade down — it simply does not count.

Small neighbourhoods

In a neighbourhood of 300 residents, a single burglary can double the statistics. We therefore pull crime rates of small neighbourhoods towards the national average (empirical-Bayes shrinkage). Neighbourhoods with fewer than 50 residents are excluded from the comparison; industrial areas and water are excluded from rankings.

Those 1,583 neighbourhoods keep their own page — reachable from the map and from neighbouring areas — but we ask search engines not to index them: a report on 30 residents or a stretch of open water does not belong in the search results.

Why there is no ‘−18% on last year’ here

Other sites put an arrow next to the crime figures: up or down on last year. We do not, and that is deliberate. We checked it. For every neighbourhood we compared the change between 2019 and 2022 with the change between 2022 and 2025. If a rise said something about a place, the two would line up. They do the opposite — the correlation is −0.18: a neighbourhood that rose last year is more likely to fall the next than to keep climbing. That holds in every size class, including neighbourhoods with more than a hundred and fifty crimes a year.

This is regression to the mean: the year-on-year change measures chance, not development. A typical neighbourhood of 500 to 1,500 residents swings 24% a year with nothing happening. Nationally there is barely any movement either: 803,626 crimes in 2019 against 808,021 in 2025. A trend arrow would look professional and mean nothing, and it would regularly contradict the grade next to it. So we show the most recent year, and say how solid that single year is.

Shopping and nightlife districts

Crimes are recorded where they happen, not at the victim’s address. In a neighbourhood with 800 residents and 520 businesses, what happens to visitors counts too. We correct for that with the business weighting, but it does not solve everything — which is why such neighbourhoods carry a note under the safety block. We never exclude them from a list.

What this is not

A low grade does not mean a neighbourhood is unliveable, and a 9+ doesn’t mean you’ll be happy there. The weighting is a choice — a reasonable one, but a choice. Prefer urban buzz over quiet? Use Your mix on the map.

Sources

Six sources, each with its own vintage and refresh rhythm. None of them is estimated or filled in by us — what CBS does not publish is listed above under ‘What we do not measure’.

Figures suppressed by CBS for privacy reasons (too few observations) are not counted. Last data update: July 2026. Last model change: 20 August 2026 (Quiet & nature: noise, greenery and trees added).

Log of changes

Every change to the model, newest first. Open one to see what changed and how far the grades moved.