New Zealand’s 2020 cannabis referendum: a historical guide

New Zealand voted on a proposed cannabis law alongside the general election on 17 October 2020. The official result was 50.7% against and 48.4% in favour, with 0.9% recorded as informal votes. Those figures describe a particular national decision in 2020, not a survey of public opinion today.

This historical guide follows the question, the counting timetable and the published result. It also explains why a vote on a proposed bill was different from putting a completed law into operation. It does not assess present-day cannabis law or provide advice about personal activities.

The question on the ballot

The Electoral Commission’s 2020 results record preserves the question: “Do you support the proposed Cannabis Legalisation and Control Bill?” The two answers expressed support for, or opposition to, that proposal.

The reference to a proposed bill was important. Voters were considering a particular policy framework, rather than choosing between every possible approach to cannabis policy. A rejection of that proposal did not reveal which alternative each opponent preferred. Equally, a supporting vote did not separately measure agreement with every clause.

The ballot therefore recorded a decision, not an explanation of each voter’s reasons. The totals cannot establish whether someone was motivated by a particular provision, their broader policy views or another consideration. Those questions would require different evidence from the ballot count itself.

A proposal, not an already enacted framework

The Ministry of Justice published an explanatory guide to the exposure draft. Its chapters covered a proposed regulator, licensing, product controls, packaging, retail premises, advertising and enforcement. This was a connected regulatory model, not simply a single change of wording.

The guide described reducing harm as an objective of the proposed framework. That was an aim stated in the proposal, not evidence that the proposed system had already operated or achieved its intended outcomes. Describing an objective accurately does not establish that supporters’ expectations or opponents’ concerns would necessarily have been realised.

The guide also presented the legislative step conditionally: a bill could be introduced after the election and referendum, with an opportunity for public input. Its explanations concerned what the draft contemplated. Reading it as an instruction manual for an existing scheme would confuse the document’s historical purpose.

Why the referendum was non-binding

Before the vote, Parliament’s explanation dated 21 September 2020 expressly described the cannabis referendum as non-binding. The result would not, by itself, enact the proposed framework. The Government retained the decision about introducing legislation.

Parliament explained that any introduced bill would still undergo the normal parliamentary stages, including consideration by a select committee and an opportunity for public submissions. A supportive result would therefore have been one part of a possible process, not the completed process.

This distinction also avoids an overstatement about the negative result. The count showed that the proposal did not obtain majority support. It was not itself the text of a new Act. A historical account can report the vote’s outcome without turning that event into a comprehensive explanation of either earlier or later law.

The voting and counting timetable

In its 16 October announcement, the Electoral Commission identified the following day, Saturday 17 October, as the final opportunity to vote in the election and referendums. It also explained that the referendum papers would not be counted on election night: priority went to the general election results.

That announcement distinguished three events that are sometimes compressed into one date in later summaries:

• 17 October 2020: election day and the close of voting.
• 30 October 2020: publication of preliminary referendum results.
• 6 November 2020: publication of official results, including special votes.

A report dated on polling day could not contain the subsequent official referendum count. Keeping the timetable visible makes older headlines easier to interpret and helps distinguish an early report from the final record.

What the preliminary count showed

The Commission’s 30 October release reported 46.1% in favour and 53.1% against. It explicitly described these as preliminary figures based on ordinary votes cast in advance and on election day. Special declaration votes were not included.

Those percentages belong to that stage of the count. They should not be presented beside final totals as though all the figures came from one table. The release itself told readers that the official result would follow and would include the special votes.

The distinction is straightforward but useful: an early count can accurately describe the papers counted so far without being the completed result. A later percentage can differ because the later announcement covers a different stage of the process. The two dated releases should be read in sequence, not treated as rival final accounts.

The official result and its denominator

On 6 November, the Commission released the official results. The cannabis referendum table recorded:

• Yes: 1,406,973 votes, or 48.4%.
• No: 1,474,635 votes, or 50.7%.
• Informal: 26,463 votes, or 0.9%.
• Total: 2,908,071 votes.

The release confirmed that special votes were included. It described informal papers as unmarked papers or papers on which the voter’s intention was unclear. It also noted that voters could participate in both referendums, only one, or neither.

For this reason, the informal category must not be silently described as another group of opponents or supporters. Nor do these percentages measure the views of every New Zealander. They describe the referendum count using the categories and total stated in the official table.

The final gap between the two answers was 67,662 votes, calculated by subtracting the yes total from the no total. This arithmetic describes the margin; it does not supply an explanation for individual voting decisions.

Why some accounts quote 51.2% against

The Parliamentary Library’s historical election analysis reports 51.2% against when discussing valid votes. The Commission’s 50.7% figure uses the larger total that also includes informal votes. The difference is the denominator, not a different winner.

Using the published totals, the valid-answer denominator is the sum of yes and no votes. Adding the informal category produces the overall total shown in the Commission’s table. A percentage calculated from one denominator should not be compared directly with a percentage from the other without explaining that change.

This is why retaining the label beside a percentage matters. “Of all votes recorded” and “of valid votes” are different statements. An account that quotes a percentage accurately but drops its denominator can still leave readers with a mistaken impression about the count.

What the official information programme was for

The Ministry’s account of its public information programme explains that it prepared materials on both referendum topics. The stated approach was to provide factual, impartial explanations, using language close to the relevant Act and proposed bill while making technical material more accessible.

That role was different from campaigning for either answer. The Ministry says the explanatory programme was not intended to debate the merits of the proposals. Its documents are consequently useful evidence of what voters were being asked to consider, rather than proof of what every voter understood or why they voted as they did.

The Ministry also identifies the former referendum website as an archived resource whose information concerned the 2020 election. An archived page can preserve an important explanation while remaining tied to its original date. Its historical value does not turn its statements into an assessment of circumstances today.

Reading the record without adding claims it cannot support

The strongest account of this referendum keeps three kinds of evidence separate: the proposed framework, the dated announcements during counting, and the completed results. Each answers a different question. A proposal explains an intended model; a counting announcement explains progress; the result records the choices counted.

The same discipline applies to interpretation. The national totals demonstrate substantial support on both sides and a larger no vote. They do not, on their own, establish how a different proposal would have performed, how opinion later changed, or what any future government would decide.

The essential historical conclusion is limited and clear: the proposed bill did not receive majority support in the official 2020 count. Understanding that outcome requires the actual question, the final rather than preliminary figures, and the distinction between a referendum result and the parliamentary process described at the time.

Sources

These official historical sources were checked on 13 September 2026. The dates discussed in the article belong to the 2020 vote and its records.

• Electoral Commission: 2020 referendum results.
• Electoral Commission: final voting-day announcement, 16 October 2020.
• Electoral Commission: preliminary results, 30 October 2020.
• Electoral Commission: official results, 6 November 2020.
• New Zealand Parliament: explanation before the cannabis referendum.
• Parliamentary Library: analysis of the 2020 election and referendums.
• Ministry of Justice: guide to the exposure draft.
• Ministry of Justice: the 2020 public information programme.