Published guidance says a New Zealand court usually signs an urgent temporary protection order within 24 to 48 hours, and gives no timeframe for the delivery step that a conviction for breaching that order generally depends on

What is happening?

Under the Family Violence Act 2018 a New Zealand court can make a temporary protection order without notice, which means the order is made before the person it is made against has been told that an application exists. The order takes legal effect the moment a judge signs it.

Enforcement does not begin at the same moment. For a conviction for breaching the order, the general requirement is that the order has first been served on the respondent, meaning physically delivered to them. Where the respondent already knew about the order by some other route, that knowledge can stand in place of service. Arrest before service remains possible. A conviction generally does not.

Delivery is not one job held by one body. It may be carried out by court bailiffs, private process servers or police officers.

The court stage carries published timings. An urgent without-notice application is usually decided within 24 to 48 hours, and an ordinary application is placed in front of a judge within two working days of filing. The published material that carries those court timings also names who delivers the order, and no source opened here gives a time for that delivery.

Breach is a criminal offence that is charged and prosecuted. Ministry of Justice figures presented through a data publishing service record breach charges ending in conviction at 4,076 in the financial year to June 2020, 3,275 in the year to June 2022 and 3,735 in the year to June 2024. Charges ending as not proved over the same three years were 1,496, then 1,177, then 1,422.

Whose problem is this?

No single body in that list holds the whole step. The court decides the order and the police may or may not be the party that delivers it, and the statute attaches the consequence to delivery rather than to the decision.

Where does this problem end?

The boundary here is narrow on purpose. Nothing in this document says that an order should not have been made or that it should have contained different conditions. What it describes is the one step between a decision that has already been taken and the point at which that decision can be enforced in the ordinary way.

What is the state now, and what should it be?

Now

Needs a new measurementthe target state: no source opened here names a target time for delivering a protection order, a performance standard for the agencies that deliver them, or a statutory deadline. The Act itself could not be opened in this round, so the absence recorded above is an absence in the guidance and statistics that were opened rather than a confirmed absence in law. Public commentary about the delay uses the framing that any wait is too long, and that is an advocacy position rather than an official target, so it is not adopted here as the destination. The destination in this document is therefore recorded as absent rather than filled in from elsewhere.

How big is it?

The number of people standing in the delivery interval was not established here, and the affected-population module for this document records the count as not derivable. No source opened here gives the annual number of temporary protection orders issued or the annual number served. The Ministry of Justice publishes annual data tables and the figures sit inside downloadable spreadsheets that were not opened in this round, so this is a limit of the round rather than an absence of data.

What is counted is the volume of breach proceedings that follow. Charges for breach of a protection order ended in conviction 4,076 times in the financial year to June 2020, 3,275 times in the year to June 2022 and 3,735 times in the year to June 2024. Over the same three years, charges recorded as not proved numbered 1,496, then 1,177, then 1,422.

Those figures count charges rather than people, and they count what happened after an order was already in force and already delivered or otherwise known about. No source opened here states how many charges turned on proof of delivery, and this document does not perform that subtraction. Reading the not-proved column as a measure of delivery delay would be an inference no opened source supports.

The surrounding caseload is large and is counted on a different basis again. Police family harm investigations were recorded at 16,361 in December 2023 and 13,618 in July 2024, and a general reference compilation records that in 2020 police attended a family violence call about once every four minutes. Neither figure is a measure of delivery, and no opened source links either of them to how long delivery takes.

Under what conditions does it arise?

1. The order is designed to bind before the respondent knows about it. An order made without notice exists precisely because telling the respondent in advance can be unsafe. The same design creates an interval in which the order is in force and the person bound by it has not yet been handed it.

2. Legal force and enforceability attach to different events. Force attaches to the signature. The general route to a conviction attaches to the delivery. Between the two events the order is fully real in law and thin in the ordinary enforcement sense, and nothing in the sources opened here sets how long that state may last.

3. The step is shared among three kinds of agent. Court bailiffs, private process servers and police officers may each carry it out. Where more than one party may act and no rule states by when, a slow delivery is not a breach of anything.

4. One stage of the sequence is published with a timing and the next is not. The government page that names the delivery agents also states that an urgent application is usually decided within 24 to 48 hours and that an ordinary application reaches a judge within two working days. No equivalent figure for delivery appears in any source opened here.

5. Elapsed time produces no event. No source opened here describes an automatic review, a notification to the protected person, or a published series that starts counting when the judge signs. A quantity nobody is required to reach cannot be missed.

What has been tried?

Three of these four are real changes to the family justice system and none of them attaches a time to the step this document describes. Two of them move the court stage, which already has a published timing, and the third changes what a court can order rather than how quickly an order reaches the person bound by it.

What was found?

Why is it still unsolved?

Enforcement absent — the order is in force from the moment it is signed, and until it has been delivered the ordinary route to a conviction for breaching it is not open.

The first part is that two things which read as one thing are attached to different events. Signature creates the obligation and delivery creates the proof. In the interval between them the order is entirely real, the respondent is bound by it, arrest remains available, and the general path to a conviction is not yet there. That interval is closed by an operational act rather than by a rule, so its length is decided by whoever is free to carry the paper.

The second part is that the sequence is published with a timing for the stage that ends in a signature and no timing for the stage that ends in a delivery. The court side is stated plainly: an urgent application is usually decided within 24 to 48 hours and an ordinary application reaches a judge within two working days. In the sources opened here the delivery side is described only by who may carry it out. A reader of that material can therefore find out how quickly the order will exist and cannot find out how quickly it will arrive.

The third part is that no single body holds the step. Court bailiffs, private process servers and police officers may each serve an order. Where several parties may act and none of them is given a deadline, a slow delivery is not a failure against any published commitment, and the question of who should have moved faster has no answer inside the documents that describe the process.

The fourth part is that the interval generates nothing. It produces no automatic review when days pass, no notification to the person the order protects, and no routine public series. The only figures identified in this round came from an official information request rather than from a published statistic.

What observation would mean it is solved?

Candidates — (a) a published timeframe or a statutory deadline attaches to the delivery of a protection order and takes effect (b) a routine public series reports how long delivery takes, updated each year and broken down by region (c) the interval between signature and delivery falls and stays down.

(a) alone counts paper. A timeframe set above the time delivery already takes changes a document and not a person. A deadline can also carry no consequence for missing it, which is the position the delivery step is already in, and adding a number without attaching anything to it reproduces the same state with a figure printed next to it.

(b) alone measures without bounding. A published series would make the interval visible and would not shorten it. It is still the more load-bearing of the three, because nothing in the other two can be checked while the quantity itself is unpublished. A series would also have to state its counting basis, since an average taken across deliveries that completed will behave differently from one taken across all orders in force.

(c) alone is not observable today. There is no published baseline against which a fall could be measured, so this candidate depends on (b) existing first. It would also move with the number of orders made, with how many respondents are easy to locate, and with which of the three kinds of agent carries the task, none of which is a change in the design this document describes. The three have to be read together, and (a) should not be read as progress unless (b) exists to show what it did.

What is it connected to?

Fills with researchcourt processing time and the on-notice hearing stage that follows a temporary order, police workload and deployment for family violence work, the review of family violence deaths and whether it examines the timing of protective measures, the operation of police safety orders as a short-term alternative, and the service of protective orders wherever else it is measured. Relation type and evidence grade were not confirmed in this round.

What these sources do not say

• How long delivery actually takes. A 2026 report based on an official information request is described as giving national and regional average times for police to deliver temporary protection orders across 2021 to 2025. That page refused automated access in this round and no source opened here repeats its figures, so none of them is carried in this document and no statement above rests on them. They are recorded as identified and not carried.
• The statutory text. The New Zealand legislation site refused automated access, so every statement here about when an order takes effect and what a conviction requires rests on published guidance from a legal information manual, the Ministry of Justice, New Zealand Police and a government victims information service. No section number of the Family Violence Act 2018 is cited anywhere in this document.
• The maximum penalty for breach. No source opened here supplies it, so this document describes breach as a criminal offence that is charged and prosecuted and gives no penalty figure.
• The annual number of protection orders. The Ministry of Justice publishes annual data tables and the figures sit inside downloadable spreadsheets that were not opened here. This is a limit of the round rather than a statement that the count does not exist.
• Whether delivery timing affects the outcome of breach charges. Charges ending as not proved numbered 1,496 in the year to June 2020, 1,177 in the year to June 2022 and 1,422 in the year to June 2024, and no opened source states how many of them turned on proof of service. No opened source performs that subtraction and neither does this document.
• Whether delay in delivery has been linked to later harm. A national committee reviews family violence deaths and publishes reports. No report opened here addresses the timing of delivering protective orders, so this document makes no claim in either direction about that link.
• A published response from New Zealand Police. No statement by New Zealand Police about the time taken to deliver protection orders was found in this round. The report that would have carried any such response could not be opened, so it is not known whether one appears in it.
• How the delivery task is staffed. No opened source gives how many officers or bailiffs are assigned to delivering orders, how that task is prioritised against other duties, or how the choice among the three kinds of agent is made in a given case.
• Anything after 2025. The most recent year for which any delivery figure was identified is 2025, and that figure is not carried here.

See the evidence

No primary legal document was read in full. Twelve pages were opened directly and carry everything asserted above: a community legal information manual for the rule on when an order takes effect and what a conviction requires, the New Zealand Police page for who delivers an order and how fast the court stage runs, the Ministry of Justice and government victims pages for what a protection order is, a chart presenting Ministry of Justice statistics for breach charge outcomes, the ministry data tables index and a national index of government family violence data sources for the absence of a published delivery series, a national newspaper for police family harm investigation counts and for the wider court backlog that this document excludes, the ministry news release for the 2026 legislation on litigation abuse, and a law firm commentary for the case management rollout schedule. Three sources are listed with no URL. Two of them refused automated retrieval and one was not located as a page. The most consequential of the three is the 2026 report carrying the delivery times, and because no opened source repeats those figures they are recorded as identified and not carried rather than reported with a caveat. That decision is why this document states the shape of the interval and not its length. Where the opened sources overlap they agree: the requirement that an order be served before a conviction appears in the legal information manual and the role of police as one of the delivery agents appears on the police page, and no time for delivery appears in any source opened here. On the institutional side, no published statement by New Zealand Police or by the Ministry of Justice about the time taken to deliver protection orders was found in this round, and that absence is recorded rather than filled. This is a Path A output, research-based definition, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range Not derivable

The reason and what is missing are listed under “What is missing” below