On the edge of the Barrier Ranges, Broken Hill spent the summer of 2018–19 watching a lake nearly two hundred kilometres away decide its future. The city's water had come from the Menindee Lakes on the Darling-Baaka River for more than a century, but three years of drought and heavy upstream extraction had drained the lakes to a sliver of their capacity, and by late 2018 a city of around 18,000 people (approx.) was staring at the real prospect of running out. Then, in January 2019, downstream at Menindee itself, the crisis turned visible and grotesque: a blue-green algae bloom collapsed after a cold snap, stripping oxygen from the weir pool, and up to a million fish died over a matter of days (approx.), among them Murray cod believed to be a hundred years old. Locals waded into water thick with rotting carcasses; the smell carried for kilometres. Broken Hill's near-miss and Menindee's fish kill were the same failure seen from two ends of the same river — a system so over-committed that neither the town at its end nor the fish within it had enough water left to absorb an ordinary drought.
What the government did
The Murray-Darling Basin Plan, made law by the Commonwealth in 2012, set out to fix that arithmetic: a legislated Sustainable Diversion Limit reducing water extracted for irrigation across the whole basin by 2,750 gigalitres a year on average, with a further 450 gigalitres targeted through water-saving infrastructure rather than outright cuts to farms (Murray-Darling Basin Authority, Basin Plan 2012). New South Wales holds the largest and most contested share of that task. It contains most of the northern basin — the Gwydir, Border Rivers, Macquarie, Namoi and Barwon-Darling valleys — where the biggest unresolved problem was not river extraction but floodplain harvesting: the capture of overland flood water by large irrigation properties, legal in principle but for decades neither licensed nor metered. A 2017 independent investigation by Ken Matthews AO, commissioned after ABC's Four Corners alleged water theft and political interference in the northern basin, found compliance and metering had effectively collapsed; the state government created the Natural Resources Access Regulator (NRAR) in 2018 in response. It took until the Water Management (General) Amendment (Floodplain Harvesting) Regulation 2022 — after an earlier 2020 version was disallowed by the NSW Legislative Council — for floodplain harvesting to be licensed and metered at all.
The Minns government, elected in March 2023, inherited that unfinished regulatory build and has since finished issuing floodplain harvesting licences across the northern valleys, while signing NSW up to the Commonwealth's Restoring Our Rivers Act 2023, which extended the Basin Plan's completion date — originally 2019, then repeatedly pushed back — to December 2027 (approx.) and removed the legislated cap that had constrained voluntary water buybacks since 2018. On town water, NSW's Safe and Secure Water Program has funded well over a billion dollars (approx.) of pipelines, bores and treatment upgrades for regional towns since the last drought, the flagship being the pipeline connecting Broken Hill directly to the Murray River near Wentworth — roughly 270 kilometres of pipe (approx.), finished in late 2019 at a cost in the hundreds of millions (approx.) — so the city's supply no longer depends solely on the Menindee Lakes.
Before weighing the government's record, the case irrigators make deserves a fair hearing. The northern basin's cotton, grain and horticultural districts, and the Riverina's rice and wine industries further south, generate several billion dollars of regional output and thousands of jobs a year (approx.; ABS regional data; NSW Irrigators' Council estimates), often in towns with few other employers. Water bought back from willing sellers does not disappear from the ledger evenly — it can hollow out the fixed costs an irrigation district shares between fewer remaining users, a "Swiss cheese" effect written explicitly into the Commonwealth's 2018 socio-economic neutrality test for further buybacks. Irrigators also argue, with some justice, that formal licensing of floodplain harvesting — even at generous volumes — is genuinely better than the unmetered free-for-all it replaced, because a licensed, capped entitlement can be tightened later, while an unmeasured one cannot be reformed at all.
The strongest case
The basin was over-allocated for decades, and NSW was the hardest place to fix it
The government's underlying justification is that the Basin Plan responds to a real and long-documented ecological collapse, not a bureaucratic preference. The Murray-Darling Basin Authority's own river health monitoring has put native fish populations at roughly a tenth of pre-European settlement levels across the basin (approx.; MDBA, Sustainable Rivers Audit), and the northern basin — NSW's largest share of the task — was also the least regulated, meaning the state had to build compliance infrastructure from close to nothing rather than simply tighten an existing system. Creating NRAR, running two attempts at a floodplain harvesting regulation until one survived parliamentary scrutiny, and finally metering water that had never been measured are not trivial administrative steps; other basin states with smaller, older irrigation districts had far less ground to cover. On that reading, the sequence — investigation, new regulator, disallowed regulation, redrafted regulation, licences finally issued in 2023–24 — is not delay for its own sake but the genuine difficulty of re-regulating a resource that generations of state governments, of both political colours, had allowed to go unmeasured.
The evidence
Measured against the Basin Plan's own targets, NSW's task is running well behind schedule. Basin-wide, roughly 2,100 gigalitres of the legislated 2,750-gigalitre reduction had been recovered by the mid-2020s (approx.; Murray-Darling Basin Authority, Basin Plan progress reporting), leaving a shortfall of several hundred gigalitres years after the plan's original 2019 deadline. The supplementary 450-gigalitre efficiency measures target — water meant to come from infrastructure upgrades rather than cuts to farms — has fared far worse: only a small fraction, in the tens of gigalitres, had been delivered by the same point (approx.; MDBA reporting), a gap the 2023 deadline extension exists specifically to close. NSW's own compliance record adds to the doubt. When floodplain harvesting licences were finally issued for the northern valleys, the volumes were calculated largely on historical extraction, which independent modelling cited by the Natural Resources Commission and environmental groups (secondary sources) argues locks in past over-use rather than reducing it to the sustainable limit the Basin Plan set out to achieve.
None of that shortfall is unique to NSW — every basin state has missed elements of the plan — but NSW carries the largest share of both the remaining recovery task and the northern basin's compliance failures, and it is the only state where a major extraction pathway went completely unmetered for the plan's first decade. The Productivity Commission's periodic review of the Basin Plan has repeatedly flagged that the northern basin's supply and constraint projects, several of them in NSW, are running years behind the timetable the states themselves set (approx.; Productivity Commission, Murray-Darling Basin Plan review). Each delay pushes the reckoning further into a future where climate projections suggest droughts will be more frequent and more severe, meaning the margin for error the state is banking on now is likely to be smaller, not larger, when it is tested again.
The critique
The pattern across floodplain harvesting, buybacks and fish kills is the same: NSW has consistently chosen to regulate only once the alternative became politically unavoidable, and even then to regulate around existing extraction rather than to genuinely reduce it. Five years passed between the Matthews investigation exposing the compliance collapse and a floodplain harvesting regulation actually surviving parliamentary disallowance — five years in which the state's most over-allocated valleys operated, in effect, on trust. When licences finally arrived, the government's approach of basing entitlements on historical use meant the reform legalised and capped extraction close to where it already sat, rather than pulling it down toward the sustainable limit the Basin Plan exists to enforce; that may be a defensible first step, as the steelman case argues, but it is not the same as fixing the underlying over-allocation, and the government has not published modelling showing when, if ever, the northern basin's licensed volumes will be brought into line with the plan's environmental targets.
The clearest evidence that operational fixes have not kept pace with the underlying problem is that Menindee's fish kill was not a one-off. In March 2023, within weeks of the Minns government taking office, a second mass die-off hit the same weir pool — an estimated hundreds of thousands of fish (approx.), following a similar sequence of flood, algal bloom and low-oxygen "blackwater" after a heatwave. An independent expert panel commissioned jointly by the NSW and Commonwealth governments after the 2019 event had already recommended aeration infrastructure and real-time oxygen monitoring for exactly this scenario; the fact that a comparable event recurred within four years shows those recommendations were not fully in place in time — a gap that sits with the government in office when the second kill occurred, whatever the drought and flood conditions it did not create.
Broken Hill was saved by a pipeline built after the water nearly ran out; the fish at Menindee were not saved twice.
A pattern of fixing towns after the crisis, not before
Broken Hill's pipeline is the state's proof that it can act — but it is also proof of when it tends to act. The project had been discussed for years before the 2018–19 drought forced it into an emergency build; NSW only committed the funding and fast-tracked construction once the Menindee Lakes were close to empty and the alternative was trucking water into a city of 18,000 people (approx.). The same reactive pattern shows up further afield: smaller regional towns, including Walgett, Bourke, Guyra and Murrurundi, moved onto carted or heavily restricted water at points during the 2017–20 drought before infrastructure fixes followed, several of them years after their vulnerability had already been identified in state water security planning documents (approx.; NSW Department of Planning and Environment, regional town water risk assessments). The Safe and Secure Water Program has since worked through a substantial list of upgrades, but the state's own reporting shows a pipeline of identified, unfunded projects still sitting behind the completed ones, meaning some regional towns remain one bad drought away from repeating 2019 (approx.; NSW Budget Papers, water infrastructure programs).
The risk to the future
A 2027 deadline the state is not yet on track to meet
If NSW does not close the gap on both the Basin Plan's water recovery target and its own regional town water pipeline before the next serious drought, the state risks repeating 2019 on two fronts at once: another environmental collapse on the Barwon-Darling, and another town discovering its water security plan was still a draft when the river ran low. Climate projections used in the state's own water planning point to hotter, drier conditions and more volatile rainfall across the basin's NSW catchments, which shortens the window between a manageable dry spell and an emergency for both farms and towns (approx.; NSW Department of Climate Change, Energy, the Environment and Water; CSIRO). Floodplain harvesting licences calculated on historical extraction, rather than on the volumes the Basin Plan's ecological targets actually require, leave the state exposed to a widening credibility gap with downstream South Australia and with the First Nations communities of the Barwon-Darling, who have twice now watched the same river die in the same place. The Minns government did not create the over-allocation it inherited, but every year it manages the licensing regime without visibly narrowing that gap is a year it owns the outcome, not just the paperwork.