How we check dog rules

The promise

Every walk listed on this site carries a verdict we can trace. For 311 of them a named rule-setter has stated the rule and we tell you who. The rest sit on licensed public walk data published on the conservative reading, with the authority that governs the spot named on the page so you can check it at the source. That promise is deliberately narrow, and it is why we would rather list fewer walks than guess. We map 1,776 walks across 15 regions, and that number would be considerably larger if a missing answer counted as a yes.

It does not. Absence of evidence is published as unknown, and an unknown is not listed as dog-friendly. The reason is that the two possible mistakes do not cost the same. Tell someone a walk is closed to dogs when it is open and they miss a good morning. Tell someone a walk is open when it is closed and they can end up standing on land where having a dog there is an offence: on conviction, up to 12 months imprisonment or a fine of up to $10,000 under Conservation Act 1987 s 26ZZP or National Parks Act 1980 s 56I, with the dog liable to be seized and impounded. Those outcomes are not symmetrical, so we do not treat the two errors as if they were. Where the evidence supports both a permissive and a restrictive reading, the restrictive one wins and stays until something authoritative says otherwise.

The order we read sources in

For each place we walk an ordered list of sources and stop at the first one that yields an actual designation. The verdict is stored with its citation, because a verdict without a citation is not publishable.

1. A mapped council dog zone. Dog bylaws are written and enforced by city and district councils, not regional councils, so where a walk falls inside a zone we hold, that zone governs. We hold 2,016 zones from 13 councils, 73 of them carrying seasonal windows with real dates and hours that flip over summer., 73 of them carrying seasonal windows with real dates and hours that flip over summer.
2. DOC's own dog-access statement for the place. On public conservation land the answer lives on the DOC page for the specific track, hut or campsite. DOC runs a three-state model: places approved for dog access, places that additionally need a DOC dog permit, and everything else, which is closed to dogs. There is no blanket permit covering conservation land, and DOC sets no lead length. Lead rules are set place by place, with an "under control" class where no lead is required, so we read each place on its own terms. See DOC on where you can take your dog.
3. Coded track data. DOC's per-track dog field and Herenga ā Nuku track records, used where neither of the first two speaks.
4. Nothing. Then the verdict is unknown, and unknown is not a listing.

Three rules hold the stack together. On conflict, the stricter or better-sourced reading wins, and when we collapse two records of the same walk the surviving entry inherits the more authoritative lane's verdict rather than the friendlier one. A reader report never sets a verdict by itself: it triggers a re-check against the source, and the source decides. And national parks are closed to dogs by statute unless DOC has issued a permit for that park, which it does not do for pet walking, with certified disability assist dogs the standing exception, since they may accompany their handler anywhere on conservation land.

What each verdict actually rests on

Aggregate claims about checking are easy to make and easy to break with one counter-example, so here is the split. 311 of the 1,776 walks we list carry a verdict traced to a named authority: DOC, Herenga ā Nuku, or a council dog zone. For the remainder we publish what the base data supports: on-lead for 1,334 of them and off-lead for 131, and the walk page tells you which authority governs that spot so you can check it at the source., and the walk page tells you which authority governs that spot so you can check it at the source. On-lead is the conservative default: it is what we publish when we know where a walk is but not what the rule-setter has said about dogs.

Off-lead is the most permissive claim we make, so it gets the sharpest disclosure. Of the 287 walks on our off-lead list, 131 rest on that base lane rather than a named authority, and we are working through them, confirming or demoting each one against a council zone or a DOC field. Council resolution is honest about its own reach too: the live map resolves your position against all 2,184 zones as you move, while a static walk page only carries a council rule where the walk falls inside a zone we have mapped.

The same discipline runs through the overnight data. Of the 527 DOC huts we map, 398 have no stated dog access and are labelled that way rather than being assigned a permit requirement nobody has declared. Of the 78 DOC campsites, 45 are lead-only with no permit at all, 28 need a DOC dog permit before you arrive, 4 state nothing about dogs, so we do not count them as welcoming, and 1 carries its own conditions we spell out on the campsite page.

What we refused to publish

The most useful proof that a site is not guessing is the list of things it declined to say. Each build records its own exclusions, and this is the current tally.

• 20 walks unlisted for no dog evidence. Real routes on conservation land where no source states a dog rule. Publishing them would have meant inventing one.
• 44 routes unlisted as no-dogs or unknown. Where the rule-setter says no, the walk does not go on a dog-walking site. Some of these still appear in where dogs cannot go, because knowing the answer is no is worth something.
• 19 unverified routes unlisted. Records we could not tie back to a source we trust.
• 59 duplicates collapsed. The same walk arriving twice under two names, merged to the better-sourced verdict.
• 7 walks pulled for corrupt geometry. Routes whose coordinates put them somewhere they are not. A walk we cannot place is a walk whose zone we cannot check.

The no-evidence set is the one worth dwelling on. Had we published it on a sensible-looking default, most of those walks would have gone up as off-lead and the rest as on-lead, and one of them, Ahuriri Estuary, is recorded by DOC as no dogs. One bad default, one wrong answer, on a page a reader would have had every reason to believe. That is the entire argument for treating no evidence as no listing.

Where the rules come from

The register below is generated from the live rule layer when the site builds, so it cannot drift away from what the site is actually using. Each row names the source, how many zones we hold from it, when it last refreshed, and any caveat that travels with it. The licence for every source is in the full attribution below. Alongside the councils we hold 168 DOC polygons where dogs are prohibited outright, which is what lets a walk be flagged as closed rather than merely unlisted.

Two caveats travel with that table rather than being tidied out of it. Auckland's zones come from a 2024 snapshot rather than a live feed, so treat them as a guide and let the signs on the day decide. A handful of other councils reach us as curated extracts of a dated bylaw, not as a maintained data service, and the register says which and from when. None of these sources were built for us, and none of them are endorsements: they are public data we read carefully and cite.

Underneath the data sits the law itself. Council rules trace to dog control bylaws, conservation land traces to the Conservation Act 1987 and the National Parks Act 1980, and the controlled and open dog areas those Acts refer to are declared by notice in the New Zealand Gazette. There is no national spatial dataset of dog access. It does not exist, which is why this work is done a place at a time.

How often each thing updates

Currency claims are cheap, so here is the actual schedule, in three tiers, matching the jobs that run.

• Nightly: council dog zones, DOC alerts, and the hazard layers (pesticide operations, track closures, kauri dieback areas). These are the layers most likely to change under you, and they are the ones a stale answer hurts most.
• Weekly: track and feature geometry, and the discovery pass that finds walks we do not yet hold. New walks arrive through the same evidence stack as everything else, so a discovery does not become a listing on its own.
• Hand-checked and dated: the 32 published verdict corrections, the curated bylaw zones, and the 26 famous-track answers we maintain by hand because a wrong answer on those is the one most people would see.

We keep two dates apart on purpose. When the data last refreshed is a fact about the pipeline. When the method was last reviewed is a fact about us. Split the byline into two stamps, 'Data refreshed 25 July 2026' from live-rules.json and 'Method last reviewed 1 August 2026' set by hand, or delete the clause 'so each page carries both'.

Some rules do not move on any schedule, and we do not dress them up as if they did. Public conservation land sits in a restricted fire season all year round, not only over summer, and national parks are closed to pet dogs by statute. Those hold whatever the nightly job did. The seasonal beach windows are the opposite case: they change with the date rather than with a data refresh, so the site resolves them against today rather than describing them in general terms.

What we do not claim

A methodology is only as useful as the list of things it will not do for you.

• Signs on the day win. A sign at the trailhead, a rāhui, a closure notice, or a ranger reflects a decision made after our last refresh. If our verdict and the sign disagree, the sign is right and we want to hear about it.
• A zone is a boundary, not a blessing. Our polygons are drawn tight to the bylaw's own boundary. They tell you which rule applies where, and nothing more.
• Nobody has endorsed us. No council, and not DOC, has reviewed, approved or partnered with this site. We read their published rules and cite them, which is a different thing and we will not imply otherwise.
• An unlisted walk is not a banned walk. Most of what is missing here is missing because nobody has stated a rule we can point at, not because the answer is no. Ring the local DOC office or the council and ask.
• Verdicts are about rules, not conditions. Off-lead means it is allowed, not that it is a good idea near a road, a cliff, livestock, or nesting birds. That judgement is yours, and walking with your dog covers what we would weigh.
• No accuracy percentage. We do not publish one, because we have no way to compute it honestly. The lane split above is the closest true statement we have.

Corrections, and how to send one

Every correction we make to a published verdict is logged with the walk, what it should be, and the source that settled it. The ledger is public because a site that never publishes a correction is either perfect or not looking.

The ledger earns its keep in both directions. One entry exists because a walk crossing a national park was carrying an on-lead verdict it should never have had. Another exists because a hand-corrected no-dogs verdict silently reverted to on-lead in a later build, and the ledger is what caught it. Corrections are re-applied on every build for exactly that reason.

If something here looks wrong, tell us with the form below: it lands straight in our checking queue, and if you leave an email we will tell you what we found. Useful reports include the walk name or its link, what our page says, what you saw or were told, and where that came from: a photo of the sign, the council page, the DOC page, or the name of the office you rang. A date helps, since rules change and we would rather know which version you met.

We aim to reply within five working days. A report does not flip a verdict on its own, and we would not trust a site where it did. What it does is send us back to the source, and then one of three things happens: we correct the page and log it here, we make the wording clearer, or we come back and show you the source we relied on. If you would rather ask a question than report a problem, the FAQ and guides are the place to start.