For the decade before it was removed in 2016, the level crossing where Main Road met Furlong Road in St Albans, in Melbourne's outer west, recorded two deaths, three collisions and close to eighty near misses — boom gates coming down on cars, cyclists and pedestrians squeezed between a busy arterial road and one of Melbourne's busiest rail lines (Level Crossing Removal Project case data, cited in government and industry reporting, approx.). Multiply that pattern across dozens of similar intersections in Melbourne's growth corridors and the case for removing them stops being an abstraction. It is the reason level-crossing removal has remained one of the most popular things any Victorian government does — and why a program that started as a 50-crossing, $5–6 billion pledge has grown, under three premiers, into a 110-crossing commitment costing several times that.
The Allan Labor government did not invent that case. It inherited a program launched by Daniel Andrews before he became premier, has expanded it twice more since taking over, and now owns both its genuine safety record and the questions its own state auditor has raised about whether every dollar spent has bought a dangerous crossing genuinely worth removing.
What the government did
Daniel Andrews, then Labor opposition leader, promised to remove 50 of Melbourne's most dangerous and congested level crossings by 2022 ahead of the November 2014 election, at an estimated cost of $5–6 billion (Level Crossing Removal Project, Program Business Case Executive Summary, May 2017, approx.). In office, the cost moved quickly: the 2016 program business case put it at $7.6 billion, and by July 2017 — before a single extra crossing had been added to the list — actual project cost stood at $8.3 billion, an increase of more than 38 per cent on the original estimate (Victorian Auditor-General's Office, Managing the Level Crossing Removal Program, 14-Dec-2017).
The target then grew in stages. The government committed to 75 crossings by 2025 at the 2018 election, taking the total funded program to $14.8 billion by the time the Auditor-General checked back in 2020 — roughly $6.5 billion more than the cost recorded in 2017 (Victorian Auditor-General's Office, Follow up of Managing the Level Crossing Removal Program, 14-Oct-2020). A further ten sites were added in a mid-term budget announcement in July 2021, taking the total to 85, and the November 2022 election — the last Andrews contested as premier — added the current target of 110 crossings, originally promised by 2030. Andrews resigned in September 2023; his successor, Jacinta Allan, has continued the program under her own government. More recent government material puts full completion at 2032 rather than 2030, and in 2026, asked whether a specific 2027 sub-deadline for the Upfield Line crossings still held, the responsible minister confirmed only the broader — and by then already slipping — overall target, rather than the original date (AAP News, 2026, approx.).
Delivery sits with what is now the Major Transport Infrastructure Authority (MTIA), the state's dedicated major-projects agency, which folded the original Level Crossing Removal Authority in alongside the North East Link, West Gate Tunnel and Suburban Rail Loop as Victoria's list of megaprojects grew.
The strongest case
A known killer, methodically removed
Put at its strongest, the government's case rests on results a critic has to concede. Boom-gate crossings are not a hypothetical danger: a 1991 Victorian study found that upgrading 91 dangerous crossings from flashing lights to boom barriers between 1971 and 1989 cut collision deaths at those sites from 61 to two (cited in TrackSAFE Foundation research, approx.). The government's own figures put the removal program's effect at a similar order of magnitude: monthly train-and-vehicle collisions and near misses across the metropolitan network fell from an average of 21 in 2017 to around five in 2023, a drop of close to three-quarters, with the program credited with preventing on the order of 117 crashes and near misses a year (Victoria's Big Build / Department of Transport and Planning safety reporting, approx.).
The congestion case is just as direct: a boom gate down for a large share of the two-hour peak, banking traffic back for blocks, is now simply gone at every removed site, and many of the elevated sections have delivered new parkland, cycling paths and community space beneath the rebuilt rail line — an amenity gain residents list alongside the safety benefit. None of this is contested even by the program's harshest critics, and it shows in the polling: surveys during the most contested phase of the program found roughly six in ten Victorians supported elevated rail as a removal method, despite vocal local opposition in individual suburbs (polling cited in state and industry reporting, approx.). A program that removes a genuine, well-documented cause of death and delay, and does so with sustained public support, has cleared a bar most infrastructure projects never reach.
The evidence
The government's own auditor has tested this program twice, and neither report is comfortable reading set against the achievement above. The Auditor-General's first review, tabled in December 2017, went beyond cost growth and asked a sharper question: were the 50 selected crossings actually the most dangerous and most congested in Melbourne? Its answer was not straightforwardly yes — cross-checked against the state's own risk-ranking tools, only around six in ten of the selected sites ranked in the top 50 for danger or congestion, while several other crossings that did rank in the top 50 were left off the list (Victorian Auditor-General's Office, 14-Dec-2017). The report found the selection reflected the original election commitment rather than a rigorous, weighted options assessment, and that the business case meant to justify the sites was not finalised until April 2017 — almost two years after work had already begun on many of them.
The follow-up audit, tabled in October 2020, found genuine improvement: for the second stage of the program, the delivery agency built a transparent, four-part framework — safety, movement, place and delivery efficiency — and applied it consistently, a real fix to the process failure identified in 2017. But the same report found the deeper problem unresolved: no formal cost-benefit analysis was prepared before government approved expanding the program from 50 to 75 crossings, so a roughly $6.5 billion jump in committed spending was never tested against a quantified benefit case in the way major projects are normally required to be (Victorian Auditor-General's Office, 14-Oct-2020).
The only independent test of whether this program's sites were worth their cost was run against a fifty-crossing scope. It has since grown to a hundred and ten.
That gap has not been closed since. The target has grown twice more — to 85 crossings in 2021, then to 110 at the 2022 election — and the Auditor-General's most recent look at the program, its Major Projects Performance Reporting 2025 (tabled 4 March 2026), does not revisit the cost-benefit question at all. It records both current LXRP components — the 85-by-2025 project and the further 25 crossings funded to 2030 — at green self-assessed ratings across scope, cost, time and benefits, alongside a combined fall in total estimated investment of roughly $160 million between the 2024 and 2025 surveys. That fall, the report notes, reflects a reclassification of capital spending to operating spending under accounting standards, not a genuine saving (Victorian Auditor-General's Office, Major Projects Performance Reporting 2025, 4-Mar-2026).
The critique
Self-assessed green ratings are the delivery agency's own account of its own project, not an independent test of value for money — and this is a program whose only independent, ground-up scrutiny found real problems every time it was applied. The Auditor-General's 2017 finding that the site list did not map cleanly onto objective danger and congestion measures has never been re-tested at the current, more than doubled scope. Nor has the 2020 finding — that expansion decisions were made without a proper cost-benefit case — been revisited for the additions made since: the 85th crossing and the 110th were added to the list by the same kind of election commitment and budget decision that produced the original 50, not by the transparent four-principle process VAGO praised for stage two.
This is not a legacy problem the Allan government is managing under constraint. Three of the four expansions — the original 50, the increase to 75, and the increase to 110 — were election commitments; the fourth, the move to 85, came via a mid-term budget announcement. Andrews approved the expansions to 75, 85 and 110; Allan's government has presided over the subsequent slippage in the completion date. The timing of each expansion, clustered around elections, is read by the government's critics as evidence that later additions to the list have been shaped as much by where votes are close as by where crossings are most dangerous — a claim VAGO's reports do not directly test, because nobody has asked the state's auditor to test it at the program's current scope.
The gap with Victoria's other megaprojects is instructive. When the Suburban Rail Loop's business case was challenged, the Auditor-General and the state's Parliamentary Budget Office each ran a full, independent benefit-cost recalculation and published a number — 0.51 at Treasury's standard discount rate, well under the break-even threshold of 1.0 — that has framed public debate on that project ever since. The Level Crossing Removal Project, whose funded cost had already reached $14.8 billion by 2020 and has kept growing through two further expansions, has had no equivalent recalculation since the scope was fifty crossings. A program that costs less than the Suburban Rail Loop in total has nonetheless received far less independent scrutiny of whether its benefits still justify its price at the scope it has actually grown to.
The risk to the future
A finish line that keeps moving, on a balance sheet already stretched
The Level Crossing Removal Project's climbing cost is not happening in isolation. It sits on a Victorian balance sheet also carrying the Metro Tunnel, the West Gate Tunnel, the North East Link and the Suburban Rail Loop, each of which has recorded cost growth of its own, with net state debt projected to reach roughly $177.8 billion by June 2027 (cited in AAP News, 2026, approx.). Every dollar the removals program still needs is a dollar competing with those other commitments, and the same period that reaffirmed the 110-crossing target also saw its finish line described inconsistently — sometimes 2030, increasingly 2032 — while a specific, previously promised 2027 sub-deadline for the Upfield Line crossings went unconfirmed.
The deeper risk is an accountability gap that has now existed for roughly six years. Because the only independent, site-by-site test of value for money was run in 2017 against a fifty-crossing list, nobody outside government has verified whether the sixty-odd crossings added since — many in growth corridors and contested seats — clear the same danger and congestion bar as the first tranche, or were chosen under the improved four-principle framework VAGO endorsed for stage two and consistently applied since. If the finish line slips again, as the Upfield Line episode suggests it might, Victorians will have no independent audit trail to check the claim against, unless a future Auditor-General asks, at the program's current scope, the same question posed in 2017: are these still the most dangerous and congested crossings, at a price still worth paying.