On July 9, the Gandaki Province Assembly did something no Nepali legislature had managed since prohibition: It unanimously passed a bill licensing cannabis cultivation for medicinal and industrial purposes. Province Chief Dilliraj Bhatta authenticated it on August 3. No licence can be issued, however, until the Provincial Government makes rules under Section 31 of the bill and the ministry issues the standards beneath them. That drafting is where this policy will actually be decided.

The architecture deserves credit. Prior consent, then infrastructure, then a licence granted only after inspection. A purchase agreement with a registered buyer anchors supply to declared demand rather than speculation, which Thailand lacked when it delisted cannabis in 2022. Section 14 makes testing mandatory before harvest. Still, four main problems remain.

Ceiling and the number behind it

Section 4(4) caps tetrahydrocannabinol (THC) at 0.3 percent by dry weight. Section 4(5) then exempts cultivation under the full direction and control of the provincial government for medicinal or research purposes. Read together, these say what the Act’s title does not: Licensed private cultivation is a hemp programme, and the only lawful route to medicinal chemotypes runs through the government itself. A plant consistently below 0.3 percent is, by definition, a cannabidiol (CBD)-dominant Type III. Much of the clinical evidence for cannabinoid medicine, including the strongest findings on chronic pain, chemotherapy-induced nausea and spasticity, rests on preparations with THC well above that threshold. Purified CBD medicines are the exception, but even they sit awkwardly with the cap: In Type III plants, THC and CBD accumulate at a roughly fixed ratio near 20:1. So a crop bred for pharmaceutically useful CBD will often cross 0.3 percent THC before it is worth extracting.

The number deserves scrutiny. Nothing about 0.3 percent derives from Nepali germplasm or growing conditions. It is an arbitrary threshold, and some countries are now adopting 1 percent. Nepal’s cannabis may exceed both, but nobody has measured them. Section 4(2) restricts cultivation to prescribed varieties and Section 4(3) allows naturally occurring plants meeting prescribed standards, yet no registry exists and the chemotype of the plants in those hills is unknown.

A threshold also implies a compliant seed. Section 4(6) defers certification to the rules, but without a breeding programme supplying locally adapted cultivars that reliably meet whatever figure is chosen, growers will import genetics or fail tests. Cannabis is dioecious and wind-pollinated, and pollen travels kilometres, so imported males will introgress into wild populations regardless of intent.

Which THC and measured how?

Section 4(4) does not say which THC it means. The living plant makes almost none of the active cannabinoid compound. It accumulates tetrahydrocannabinolic acid (THCA), an acidic precursor that is not intoxicating and does not register on an assay looking only for the neutral form. Drying and heating convert it, at 0.877 of its mass. Most jurisdictions therefore regulate total THC, delta-9-THC plus 0.877 times THCA. If Gandaki means delta-9 alone, a crop can pass in the field and exceed the limit once dried. The rules must also fix a sampling protocol, a pre-harvest window, a moisture basis and an analytical method, since gas chromatography decarboxylates during analysis and liquid chromatography does not.

Two omissions follow. Cannabinoid expression is a genotype by environment interaction, so a cultivar certified in the valley may test differently at altitude. And nothing says what happens to a crop that exceeds the limit: no retest, no remediation, no negligence threshold. One hot lot ruins a farmer, and the incentive shifts towards concealment.

Mandatory test with no laboratory behind it

Section 14(1) requires testing. Section 14(2) says the province may establish a laboratory, partner with an institution, or designate one. May, not shall, and unfunded. Nepal has no accredited cannabis testing facility, and an obligation without capacity cannot be enforced. A peer-reviewed analysis of CBD products sold online found roughly 69 percent inaccurately labelled, which is accuracy in a market that has laboratories. What does this mean for Gandaki?

Nor does anything require anyone to examine the soil. Section 3(2) lists production potential, quality, security, monitoring and food security among zoning criteria, but not contamination or water use. Cannabis is a documented hyperaccumulator of lead, cadmium and arsenic, taking them up from prior land use even where nothing is applied to the crop.

Farm is licensed, supply chain is not

There is no licence category for processing, none for testing laboratories, none for research. Chain of custody, meticulous from field to transport, ends at the buyer’s gate: no packaging or labelling standard, no limit on what may be made from the biomass, no recall requirement. The United States permitted infusion into food before potency limits and child-resistant packaging existed, and paid for it in pediatric emergency visits.

The applicant screen has the same shape. Walls, guards, surveillance, storage and a five-year lease amount to a capital test few hill smallholders will pass. Yet, nothing requires experience in cannabis or a comparable crop, evidence of working capital, or any bar on selling the license on. A grower who must hold a buyer’s agreement before consent, and may switch buyers only with the director’s approval, also has one customer and no leverage over price. The criminal record bar covers the applicant only, by self-declaration, and employees are not screened. The Steering Committee under Section 16 is wholly political, admitting experts as observers with no conflict of interest provision. Section 26 shows the cost. It permits prohibition of any variety research shows harms human health, and since all cannabis can harm some people, a power triggered by any harm reaches every variety.

Way forward

First, fix the number on evidence, then publish standards. Commission a province-wide germplasm survey with cannabinoid profiling before setting a threshold, and fund breeding and certification so compliant seed exists domestically. Define total THC explicitly, with a sampling protocol, pre-harvest window, moisture basis, named method and stated measurement uncertainty, and add a retest right and a remediation pathway. Set action limits for yeast and mould, heavy metals, pesticide residues, aflatoxins and residual solvents, drawing on California’s regulations. Require soil testing at application, and accredit one laboratory whose certificate gates every batch.

Second, license the rest of the chain: processing, testing laboratories and research cultivation open to universities and hospitals under ethical review, as Canada, the UK and the US do. Define permitted product forms, require a recall procedure, and extend tracking to clones.

Third, screen for competence, not only capital. Require demonstrated experience, evidence of working capital and non-transferability for a fixed term, extend background checks to employees, and set a minimum security specification. Licence cooperatives and contract-farming groups so infrastructure can be built collectively, and set minimum contract terms so growers are not price-takers to a single buyer.

Fourth, put science and surveillance in place. Constitute a statutory technical board in agronomy, analytical chemistry, pharmacology and public health, with published advice and declared interests. Train and certify inspectors, and replace Section 26 with harm minimisation. Commission a baseline cannabis use and adverse events prevalence survey before the first licence, since Nepal’s only figures come from small, convenience samples of psychiatric outpatients and undergraduates.

None of this resolves the question underneath. The 1976 Act permits special licences to the Government of Nepal or an institution under its full supervision, not to a province, and Section 22(2) concedes as much by referring unlicensed cultivation back to federal narcotics law. Section 17(c) directs the Steering Committee to encourage processing of the harvest into medicines and consumer products, which is where the value and the jobs are, but Section 29 defers export to a federal framework that offers no pathway. Gandaki can licence a crop it cannot legally sell abroad, so the economic case rests on a decision Kathmandu has not taken. Amend the 1976 Act to create a carve-out and open a clear route. Until then, the person carrying the legal risk is a farmer standing in a field.

A threshold can be written in an afternoon; a laboratory takes years; a baseline can be measured only once; and a swamped landrace does not come back. The Act was the easy part.