The Case for
Through-Running
at Penn Station
Penn Station has 21 tracks, tunnel connections on both sides, and three railroads whose networks could run continuously through the station. The obstacle is not physical — it is governance, operational design, and political will. This page makes the affirmative case.
Terminal vs. Through-Running
The difference between these two operating models is the difference between a parking lot and a highway. In a terminal, the station is the destination. In a through-running system, the station is a stop on a continuous journey.
A train arrives from New Jersey, empties its passengers, and sits on a platform for 15 to 30 minutes while crews change and food service is restocked. Then it departs back toward New Jersey. The platform is occupied for the entire layover. Every minute of dwell time is a minute no other train can use that track.
Under this model, Penn Station operates more like a bus terminal than a railroad hub. Its tunnels connecting east and west exist, but trains don't use them to run through.
A train arrives from New Jersey. Passengers board and alight in 6 minutes. The train continues east through Penn Station to Queens or Long Island in revenue service, carrying passengers the other direction. The platform is free for the next arrival.
This is how Zürich, Berlin, Tokyo, and London's Elizabeth Line operate their busiest stations. The physical infrastructure at Penn Station already supports it. The missing pieces are operational — fleet, scheduling, governance, and yards.
Why the Terminal Model Was Built — and Why It Persists
Penn Station's terminal configuration wasn't engineered for modern regional rail. It was inherited from the era when separate railroads (Pennsylvania Railroad, Long Island Rail Road) built separate approaches that happened to share underground real estate. The absence of through-running today reflects the 1970s operating agreements that followed Penn Central's bankruptcy — not a considered judgment that terminal operations produce better outcomes. The agencies have continued the terminal model because it requires no coordination agreement, no shared dispatch, and no negotiated track access. Through-running requires all three.
Penn's Physical Track Connectivity
Penn Station is uniquely positioned for through-running. It sits between two sets of river tunnels — the North River Tunnels to New Jersey on the west, and the East River Tunnels to Long Island on the east — and connects via Harold Interlocking to the Northeast Corridor and beyond. Trains can physically pass from one network to the other today. Operating policy prevents them from doing so in revenue service.
West: North River Tunnels
Two bores (four single-track tubes) connecting Penn Station to New Jersey under the Hudson River. Used by Amtrak's Northeast Corridor and NJ Transit. Gateway Program adds two additional bores to the existing pair.
Northeast: Harold Interlocking
The junction in Queens where the Northeast Corridor splits off toward New Haven and branches extend to the Bronx via Hell Gate Bridge. Amtrak trains already use this route for through-service. The interlocking is the critical constraint for high-frequency cross-system service.
East: East River Tunnels
Four single-track tubes in two bores connecting Penn Station to Queens and Long Island, used exclusively by LIRR. Trains entering from New Jersey can reach these tunnels and proceed to Jamaica, Sunnyside Yard, or LIRR branches — but today, none do in revenue service.
East: Sunnyside Yard, Queens
The LIRR's primary coach storage and maintenance facility, located a short distance from the East River Tunnels exit in Long Island City. In a through-running model, Sunnyside becomes the eastern terminal for NJ Transit trains, replacing empty platform-to-yard moves with revenue service through Penn Station.
Existing Platform Widths
Penn Station's 21 tracks serve 11 platforms. A modern high-capacity station requires at least 35 feet on average for ADA compliance, level boarding, and rapid passenger clearing. The existing station falls short of that standard almost universally — and platform width, not track count, is the binding constraint on passenger throughput.
| Platform | Tracks Served | Nominal Width | Assessment |
|---|---|---|---|
| 1 & 2 | 1, 2, 3, 4 | 20–22 ft | Stub-ended (NJT only). Cannot support through-running due to East Side conflicts. |
| 3, 4, 5 | 5, 6, 7, 8, 9, 10 | 19–20 ft | Narrow width produces fragmented staircases, ADA failure, and extreme boarding times. Decking over every other track achieves 35-foot platforms on average across this section. |
| 6 | 11 & 12 | 17 ft ⚠ | The narrowest platform in the station — below the width of a typical sidewalk. This is not a through-running problem. It is the problem through-running solves. |
| 7 & 8 | 13, 14, 15, 16 | 20 ft | Insufficient for high-volume through-service without widening. Widening feasible in phases per 2024 feasibility study. |
| 9 | 17 | 20 ft | Narrow center platform. Currently low utilization. |
| 10 | 18 & 19 | ~20 ft ✓ | LIRR-only, feeds West Side Yard. Currently the only platform approaching modern width relative to its load. |
"Platform widening is feasible to construct with manageable construction-related service disruption. Selective column relocation along the edges of the widened existing platforms could allow the new platforms to access trains on both sides… The required demolition, trackwork, concourse and platform vertical circulation improvements could be implemented in phases, one platform at a time, minimizing disruption to train movements."Oct. 2024 WSP Feasibility Study, Section 5.2.2.5 — the same team that declared widening impossible in 2021
What Through-Running Would Require
Through-running is not free, and it is not simple. But every requirement below is a solvable engineering and governance problem, not a physical veto. The 2024 feasibility study confirmed that the station geometry works. What remains is building the system that can use it.
Platform Widening
The ReThinkNYC Design Concept 2 approach — decking over every other track — achieves an average 35-foot platform width across the station. The 2024 feasibility study confirmed this is constructable in phases, one platform at a time. Column relocation along platform edges allows boarding from both sides, meeting ADA standards.
The phased approach is critical: service disruption can be managed by suspending one side of a track pair at a time, rerouting train movements through remaining tracks while construction proceeds. The 2021 claim that this was infeasible was quietly retracted in the 2024 study.
Vertical Circulation
Widened platforms require new stair and escalator cores at platform ends, not just at mid-platform. Stub-end terminals concentrate passenger flow at one end; through-running stations distribute it across both ends, which naturally reduces crowding at any single chokepoint.
New concourse connections to 8th Avenue and the proposed western entry improvement at 33rd Street are part of the ReThinkNYC buildout. These are station improvement investments regardless of operating model — they serve terminal-mode passengers as well.
The 2021–2024 Reversal on Constructability
In 2021, the WSP White Paper on Through-Running claimed widening platforms within the existing station envelope was structurally impossible and required "continuous removal of 3 to 6 station tracks for extended time periods." In 2024, the same consulting team confirmed the opposite: phased, column-by-column construction with manageable disruption is feasible. Every structural objection from 2021 was quietly withdrawn. The engineering record is settled on this point.
The Power Compatibility Challenge
Penn Station serves two electrification standards that must be bridged for through-running. LIRR operates under 750V DC third rail. NJ Transit runs on 25kV 60Hz overhead catenary. Amtrak uses 25Hz overhead on the Northeast Corridor.
Through-running at scale requires either dual-system or tri-system equipment capable of switching between these standards mid-journey — the same challenge London's Elizabeth Line solved with purpose-built Class 345 Aventra trainsets, and that European cross-border services routinely solve with multi-voltage locomotives.
Dispatch & Operations Control
Today, Amtrak, LIRR, and NJ Transit each operate separate dispatch centers with no unified operations control. Through-running requires a joint operations center — or at minimum, a formal protocol for real-time coordination between the three dispatch systems, with agreed priorities for delays, diversions, and schedule recovery.
This is a governance and investment problem, not an engineering one. Unified dispatch centers exist at every major through-running hub in the world. The existing NJ Transit–LIRR coordination at Penn Station demonstrates that the agencies can cooperate; it just doesn't extend to through-service scheduling.
FRA Compatibility and Rolling Stock
Federal Railroad Administration (FRA) regulations require equipment operating on shared track with freight to meet higher buff-strength standards than European or Japanese commuter trains. This means purpose-built multi-system FRA-compliant equipment, not off-the-shelf European EMUs. ReThinkNYC's Regional Unified Network proposal includes a unified fleet procurement as part of its capital program. This is a cost, not an impossibility — and it is a cost that falls on the procurement schedule for all three railroads regardless of operating model, since their current fleets are aging.
Why Terminal Yards Are Central to the Operating Case
The current terminal model consumes platform time in two ways: trains sit while crews change, and trains make empty "pull-to-yard" moves that occupy tunnel slots without carrying passengers. Through-running eliminates both. The key is re-designating yards from dead-end storage to active terminals on a through-running network.
Sunnyside Yard in Long Island City, Queens, is the LIRR's primary coach storage and maintenance facility. It sits just beyond the East River Tunnel exits, a short run from Penn Station. In a through-running model, NJ Transit trains that currently terminate at Penn platforms would instead run through to Sunnyside — or continue further on LIRR branches to Jamaica and beyond.
This eliminates the empty platform layovers that today consume 15–30 minutes per train at Penn. Sunnyside expansion — additional capacity for storage, crew facilities, and light maintenance — is included in the ReThinkNYC RUN capital program.
On the New Jersey side, LIRR trains running through from Long Island would terminate at existing NJ Transit maintenance and storage facilities — at Meadowlands, North Bergen, or along NJ Transit's existing branch network. These facilities already exist; what changes is that they receive LIRR stock rather than NJT stock, following a paired-branch schedule.
The key concept: stop running trains to yards empty. Run them to revenue endpoints. Every empty yard move today is a missed opportunity for a revenue-generating trip in the opposite direction.
The agencies argued that because 193 commuter trains turn around in peak periods, through-running requires impossible yard expansions. The answer is straightforward: stop running trains to yards. Run them to Queens and New Jersey in revenue service. Sunnyside trips replace platform layovers. The network adds capacity without adding tracks.ReThinkNYC RUN Proposal — Train Movement Analysis
The Core Obstacle Is Authority, Not Engineering
Three sovereign entities operate at Penn Station with overlapping but non-unified authority. Amtrak owns the station and the track. MTA/LIRR controls eastern approaches. NJ Transit controls western operations. No single body has authority to mandate a through-running schedule, set unified dwell standards, or compel cross-agency fleet procurement. This is not a permanent condition — but resolving it requires political action that no engineering study can substitute for.
The Existing Coordination Baseline
NJ Transit and LIRR already coordinate operationally at Penn Station — trains from both agencies share platforms, follow Amtrak's master schedule, and communicate through an existing joint operations protocol. This demonstrates the governance floor for cooperation. What does not exist is the formal agreement that extends that cooperation to through-service scheduling and tri-agency fleet compatibility. The gap is political, not technical.
Peer Through-Running Systems
Every major urban rail hub operating at the capacity Penn Station aspires to runs through-service on at least its highest-demand corridors. The examples below are illustrative, not perfectly analogous — each has important differences from Penn Station's context that require honest acknowledgment.
The Operating Case
The capacity argument for through-running rests on three data sets from the October 2024 WSP Feasibility Study itself. The platform-width problem, the dwell-time problem, and the train-movement problem are not separate issues — they are three faces of the same terminal-operations failure.
Scheduled Dwell Times at Penn Station
The binding capacity constraint at Penn Station is not the number of tracks — it is how long trains sit idle. Every minute a train occupies a platform is a minute no other train can use it. The figures below are from Figure A-4 of the October 2024 feasibility study.
NYC-area terminal operations
Regional service, studied alternative
Originating/terminating
Originating/terminating
Through-running goal
London's Elizabeth Line achieves 24 trains/hour/direction on just 2 tracks with under 3-minute dwells at peak stations. With 12 tracks and a 6-minute dwell standard for commuters (10 minutes for Amtrak), Penn Station has capacity for substantially more service than agencies projected under terminal-mode assumptions.
"The relatively long dwell times required for Amtrak trains… and the requirement to turn back Empire/Hudson Line trains at Penn Station… result in demand for platform track space that exceeds the supply."Oct. 2024 WSP Feasibility Study — describing the problem that through-running solves
Peak-Hour Train Movements
The agencies argued that because 193 commuter trains turn around in peak periods, through-running requires impossible yard expansions. The data from Table A-2 of the feasibility study suggests the opposite: the current operating pattern wastes tunnel slots on empty movements that through-running converts to revenue service.
| Railroad | AM Peak Behavior | PM Peak Behavior | Through-Running Solution |
|---|---|---|---|
| LIRR | 64 turn around 36 run through (empty) 28 reverse direction |
89 depart outbound 35 pull from storage 54 reverse direction |
Pair LIRR branches with NJ Transit branches. Trains run NJ → Penn → Queens in continuous revenue service. Empty storage pulls eliminated. |
| NJ Transit | 39 turn around 21 run through (empty) |
55 depart outbound 19 pull from storage 36 reverse direction |
Pair NJ branches with LIRR/Metro-North. Sunnyside replaces platform layovers. Yard trips become revenue through-service to Long Island. |
| Amtrak | 28 operate through | 31 operate through | Reduce dwell to 10 minutes. Relocate crew changes and food restocking to line endpoints, not Penn Station platforms. |
The Cascading Delay Problem
Penn Station's terminal operations are artificially constrained by conflict movements at interlockings A, C, and JO. Eliminating turn-backs removes the conflict moves that cause cascading delays. Under current operations, a single 5-minute disruption typically affects 10 to 15 downstream trains and more than 10,000 passengers. Through-running reduces conflict moves — it does not eliminate disruption risk, but it removes the structural amplification that terminal operations impose.
The $9.5B vs. $16.7B Case
Section 5.2.3 of the 2024 report outlined "Network Investment Requirements" for through-running and concluded it requires "tens of billions systemwide." Look closely at what those costs actually include — and compare them to the cost of the alternative that was dropped from the current procurement scope.
Decking over every other track to create 35-foot ADA-compliant platforms on average. New vertical circulation. New concourse connections. This is the station work, distinct from the full RUN buildout.
Seize and demolish the block and a half south of Penn Station. Build a deep-cavern stub-end terminal. Existing platform width problems persist. Existing terminal operating model continues. Neighborhood destroyed.
The "Tens of Billions" Strawman
The "network investment requirements" the 2024 study attributed to through-running included expanding the Hell Gate Line to four tracks, grade-separating Shell Interlocking in New Rochelle, building the 5th LIRR Main Line track to Jamaica, double-tracking the Port Washington, Hempstead, and Long Beach branches, electrifying NJ Transit's Raritan Valley and Main Lines, and procuring entirely new FRA-compliant unified rolling stock. Every item on that list is a system-wide state-of-good-repair or regional expansion investment that the tri-state network needs regardless of whether Penn Station runs trains through or turns them around. Lumping the cost of modernizing the entire regional rail system onto the Penn Station through-running alternative was an accounting choice, not an engineering finding. The Penn Station through-running decision is correctly evaluated on the cost of fixing the station — approximately $9.5 billion for the ReThinkNYC Penn Station work — not on the cost of every deferred capital project across three states.
Technical Q&A
The agencies failed Design Concept 2 on "Operational Performance" — not on engineering. They evaluated ReThinkNYC's 12-track, wide-platform station using terminal-style operating assumptions: 20–30 minute Amtrak dwells, Empire and Hudson Line turn-backs consuming platform time, and no paired-branch through-service. Station geometry and operating plan were treated as a single question when they are two separate questions. The engineering record is clear: constructability passed, fire-life safety passed (with standard conditions), and track geometry passed. Only the operating performance "failed" — because the agencies tested a modern station with a 1970s operating model.
Fewer tracks with faster throughput is more productive than more tracks with slow throughput. Under terminal operations, Penn Station's 21 tracks handle roughly 1–2 trains per track per peak hour because of the long dwells between arrivals. Under through-running with 12 tracks and 6-minute dwells, utilization jumps to 4–6 trains per track per hour. The total number of train movements the station can handle rises substantially — and so does platform width, which directly governs how quickly passengers can board and alight. Dwell time is a function of platform width and crowding as much as anything else. You cannot reduce dwells without first widening the platforms.
This question cannot be answered reliably without access to the underlying model files — the train performance calculator inputs, dwell parameters, headway assumptions, and switch-timing constraints that the agencies use to produce their trains-per-hour figures. NJ Transit's March 2026 review of separate SOS model files objected to public-facing trains-per-hour figures, citing train lengths, switch timing, track-model inconsistencies, dwell-parameter violations, and delay-reporting methodology. That review is exactly why the underlying files matter — they contain the assumptions that determine whether the claimed capacity is sound or circular. The capacity analysis available publicly is not sufficient to answer this question with confidence.
Railroads claimed that if trains run through to New Jersey or Queens, they consume tunnel slots and prevent reverse-commute capacity. This is only true if branches are not paired across the network. A genuinely integrated through-running network pairs New Jersey branches with New York/Connecticut branches — so that a train running from Rahway to Jamaica carries reverse-peak commuters from Manhattan to Long Island simultaneously, in the same train, on the same run. You do not sacrifice reverse-peak capacity. You generate it in both directions simultaneously. The objection assumes a terminal-mode pairing logic that through-running is specifically designed to replace.
Transit leadership, including Andy Byford, created a real opening in August 2025 by stating he would promote through-running and was no longer interested in Penn South. But the fight now turns on the master developer contract and the operating assumptions embedded in the design program. Once a developer is selected and a design is underway, changing the operating model from terminal to through-running becomes a change order — increasingly expensive and increasingly unlikely. The critical moment is now, before design is locked. The operating requirements in the procurement documents will determine whether the selected developer designs around terminal operations or through-running readiness.
Through-running effectively doubles the productive reach of the existing regional rail network without adding tracks. For the first time, NJ Transit riders would have transfer-free access to Long Island, Queens, and the Bronx. LIRR and Metro-North riders would gain direct access to Newark, Secaucus, and New Jersey's employment corridors. The region's labor market geography would expand dramatically — jobs accessible within a reasonable commute would increase for millions of residents who today face a forced transfer at Penn Station. This is the economic case that no real-estate redevelopment around a terminal Penn Station can replicate.
The terminal model requires trains to layover at Penn Station platforms between peak service runs — consuming platform time and tunnel slots with no passengers on board. Sunnyside Yard, just beyond the East River Tunnel exits in Queens, can serve as the eastern terminal for NJ Transit through-trains — replacing platform layovers with a short revenue run. Trains arrive from New Jersey, pass through Penn Station in revenue service, continue to Sunnyside for the crew change and light servicing that currently happens at Penn platforms, then return for the next service. The platform and the tunnel slot are both freed. Sunnyside's expansion capacity, phased over time, is central to the operational feasibility of NJ Transit through-running at full buildout.