Tom Wright’s claims against Penn Station through-running, checked against the record

Wright's number counts people. Trains ran about 2 to 1. The 27,223 passengers from New Jersey rode in on 25 trains. The 2,490 going out rode 12. The railroads' October 2024 study gives the same outbound figure: “about 12 commuter trains from New Jersey turn back at Penn Station to provide reverse-peak-direction revenue service in the morning peak hour.”

Every train that arrives has to leave the platform, full or empty. RPA's April 2026 report describes how they leave across the day: “about 38% of the trains drop passengers at Penn Station and then run through to a yard”; another 18 percent run through as Amtrak trains; 44 percent turn back. The passenger ratio measures how full the departing trains are. The number of trains that arrive sets the number that must depart.

The ratio is falling. From 7 to 10 AM across all rail lines, NYMTC counted 201,007 passengers entering Manhattan below 60th Street and 19,049 leaving in 2019: 10.6 to 1. In 2024 it counted 152,422 and 18,383: 8.3 to 1. Outbound rail travel is back to 96.5 percent of its 2019 level. Inbound rail travel is back to 76 percent.

The planners expect the same. The Penn Station Master Plan's June 2020 forecast puts morning peak-hour boardings at Penn Station at 4,610 in 2019 and 16,900 in 2042 if Gateway and the Penn South expansion are built. Passengers getting off rise from 60,820 to 104,400. The ratio falls from 13.2 to 1 to 6.2 to 1. If neither is built, the forecast is 5.8 to 1 (76,900 against 13,300). For 2032 to 2042 the planners assumed outbound morning ridership would grow 1 percent a year. In the rush-hour direction they assumed one-quarter of 1 percent for the LIRR and one-eighth for NJ Transit.

Most rush-hour trains at Penn Station already go out the other side. Petra Messick of Amtrak described the move, called drop-and-go, at RPA's August 2024 forum: “It's a through-running move. And over 50% of the peak hour trains in Penn Station already use that through-running move. It's how we can operate over 1,300 daily trains at Penn.” Those trains bring passengers in and run empty to Sunnyside Yard in Queens or the West Side Yard.

RPA's April 2026 report finds that the station moves the most passengers when every train runs through, to a yard or from one: “converting all operations at Penn Station to revenue-to-non-revenue (drop-and-go) and non-revenue-to-revenue service would result in the highest passenger through-put at Penn Station.” It calls turning back the move to avoid: “Turn-backs are very time intensive and should be avoided, except for the stub end tracks where they are the only option.”

The FRA's recommended plan mixes the two. Up to four commuter trains an hour run through in each direction. Most other NJ Transit trains end at longer platforms on the south side of the station. Morning drop-and-go continues. The plan reaches 32 trains an hour under the Hudson in the morning rush.

In that study the lost slots come from the turnaround, and the study says so. A New Jersey train that continues east and turns in Queens comes back west and claims “a westbound slot in the East River tunnel that otherwise could be used by more heavily patronized trains from Long Island in the peak direction of LIRR travel.” The study names the plan that causes the loss: “a 100% through-running service at Penn Station with far-side turnbacks.” Exhibit A shows both plans.

The departing train itself takes nothing from inbound trains under the Hudson. RPA's April 2026 report describes the existing Hudson tunnels as “a single eastbound and single westbound track.” A westbound train in the morning uses the track that inbound trains cannot use.

RPA recommends the same move. Its April 2026 report proposes a “second run” in the rush hour: “a NJ TRANSIT train might drop its passengers and continue to the LIRR branch line to pick up a second run, replacing an LIRR train that would have otherwise been turned at Penn Station. This kind of exchange would eliminate the need to turn a train and would save space in the railyard.”

A train that turns around holds its platform longer. The FRA reports that NJ Transit allows 22 minutes for a train to turn at Penn Station. On the wide platform in the FRA's second plan, its model finds that passengers “can be fully alighted and boarded within 2.5 minutes.” Exhibit H compares the times.

The FRA finds the opposite at the station: “Through-running train operations, both revenue and non-revenue, at Penn Station can materially increase overall throughput by eliminating the time and operational complexity associated with turning trains within the station.” It names “conflicting train flows resulting from the mixing of terminal and through operations” as one of two things that limit service today.

The railroads' consultant agreed in public. WSP's slide at the August 2024 forum read: “100% through-running is efficient at Penn Station itself.” Foster Nichols of WSP, presenting it, called full through-running “a great idea at the station itself, and it sends the trains through the station faster.” The MTA's April 2021 white paper says: “A train running through a station generally needs less time at the platform (dwell time) than a turning train, also potentially increasing throughput.” RPA's Fourth Regional Plan says: “All service ends in Manhattan, preventing trains from traveling through from one part of the region to the other, and reducing the capacity of the system overall.”

The October 2024 study's own table credits its full through-running design with 24 more trains an hour under the Hudson. That is the highest figure among the four designs it tested. It credits its smaller through-running design with 16. It marks both “No” under “Maintains Existing Level of Bi-Directional Commuter Service?”

RPA's figure of 66 depends on how long it assumes a train stands at the platform. The report gives a through train 12 minutes: 6 for passengers to get off, 4 to get on, 2 to spare. It allows the 4 boarding minutes in the morning rush, when the Master Plan count shows 13 passengers getting off for each one getting on. On wide platforms RPA's own estimate falls to 7 minutes.

The newest study answers yes. The FRA released its Phase I report on July 15, 2026. It tested 86 scenarios, including “varying degrees of through-running commuter rail service between New York and New Jersey.” Through-running is one of four principles the study is built on. The plan it recommends “introduces limited through-running commuter service”: up to four commuter trains an hour in each direction. The plan also adds stairs and escalators, lengthens three platforms, and shortens the time trains stand at them. Together these changes reach 32 trains an hour under the Hudson in the morning rush, inside the existing station. The report does not say how much of the gain comes from through-running alone.

Wright cites that result in the same interview. He describes a new report that finds the station can go from 24 trains an hour to 32. Those are the FRA's figures. The plan that produces them runs commuter trains through the station.

The FRA's second plan supports “up to eight regional commuter through-running trains per hour in each direction.” It produced a timetable at 32 trains an hour. It then failed its first simulation because trains arrived late and because LIRR trains crowded the West Side Yard. One of the FRA's two fixes is more through-running: the crowding “could be avoided by developing a service plan with LIRR trains turning on the platform or a higher percentage of through-running commuter trains.” The FRA will test that in Phase II.

The earlier work is thinner than “studied several times” suggests. The October 2024 study cites “detailed rail operations simulations conducted collectively by WSP, Amtrak and NJ TRANSIT” for the time it takes to clear a platform. I asked Amtrak under the Freedom of Information Act for the models behind the study. Amtrak answered on February 9, 2026: “The work was led by MTA (and their consultant). We have no native files, operational models (because no simulations were performed) or calculation workbooks.” At the August 2024 forum, an audience member asked WSP about its concept for a regional metro service. WSP answered: “Nothing in this presentation was prescriptive or the result of an analysis; it was to illustrate the concept.”

RPA is not a bystander to these studies. It billed Amtrak for hosting that forum and for “extensive feedback” on Amtrak's through-running presentation. Its own report on through-running was on the list of work it proposed to Amtrak, conclusion included. The records are under “RPA and Amtrak,” below the claims.

RPA's own report says the LIRR will keep running single-level cars: “Long Island Rail Road (LIRR) is also committed to purchasing double-decker passenger cars where they make sense, but tunnel height restrictions will require single level vehicles well into the future.”

The FRA finds the delay in the station, not the cars. Its Phase I report names too few stairs and escalators as “the primary constraint to platform clearance.” It finds that adding them is “an effective strategy to reduce dwell times at Penn Station.” With up to 23 new stairs and escalators and less clutter, platforms clear and board “an estimated 2.4 minutes faster than currently, an improvement of 28 percent.”

RPA said the same in 2020. The platforms “aren't large enough,” it wrote, and “there is not enough vertical circulation (stairs, elevators, and escalators).”

Double-decker trains make short stops in a station built for them. Alon Levy, whose essay on the subject is titled “Don't Run Bilevels,” reports that rush-hour trains on Paris's RER A, a double-decker fleet, stop for “about 50-60 seconds at Les Halles.”

One city center is enough. Munich's S-Bahn gives “direct rail links connecting surrounding towns and communities with Munich city centre” over one route through the center. All eight lines share it, 30 trains an hour each way. Its builders designed it for 250,000 passengers a weekday. It carries about 840,000.

Zurich built a new line, the Durchmesserlinie, so that more trains could pass through its main station. Railway Technology describes the old terminal platforms: “reversal is still time consuming and an inefficient use of platform and track space.” In Tokyo, the Ueno–Tokyo Line joined lines that had ended on either side of the center. The Japan Railway & Transport Review reported trips 7 to 10 minutes shorter. It expected rush-hour crowding on the parallel section to fall from 197 percent of capacity to under 180.

London runs a second through line north to south. The Thameslink program raised it from 16 trains an hour to 24.

RPA made the comparison itself in 2020: “Through-running service would serve a similar function to Crossrail in London or RER service in Paris to get commuters into or through the city center with speed and efficiency.” The FRA built its study on “four principles that have been proven to maximize capacity at major rail stations around the world.” Through-running is one of them. Applied to the existing Penn Station, they produce the FRA's 32 trains an hour.

Where passengers get off does not decide where the train stops running. A train can deliver most of its riders to Penn Station and keep going.

Wright argued for the other centers at the same forum: “It's very desirable to try and improve connectivity between sub-centers for balanced regional growth.” RPA's Fourth Regional Plan faulted the commuter railroads on this point: “Reverse service into many job centers with strong growth potential, such as Bridgeport or Hicksville, is poor—and limits the ability of those downtowns to grow into major economic hubs.” RPA's June 2022 report on Gateway said that if more people work from home, “through-running service and an integrated regional rail network could be even more critical to reducing auto traffic and expanding economic opportunity when jobs will be scarcer and more dispersed throughout the region.”

Census records show the two markets moving apart. The Census Bureau links each private-sector job to the county where the worker lives. From 2019 to 2023, jobs with neither end in Manhattan grew 7.8 percent to 296,621 across 18 pairings, such as New Jersey to Brooklyn and the Bronx to Brooklyn. Jobs held in Brooklyn by New Jersey residents grew 13.6 percent. Jobs linking New Jersey and Manhattan in either direction fell 1.9 percent to 336,032.

Daytime population shows where the other centers are. Downtown Brooklyn, Long Island City, Newark, New Haven, White Plains, Stamford, and New Brunswick all gain people during the workday. Newark, New Brunswick, Stamford, and New Haven are stops on the Northeast Corridor, which runs through Penn Station. Exhibit C gives the figures.

New Brunswick to Ronkonkoma is one pair of stations. A train on that route also stops at Newark, Secaucus, Penn Station, and Jamaica. It carries every trip between them, including the trips that end at Penn Station.

Wright opened RPA's August 2024 forum by listing “better connections between suburban communities” among RPA's priorities for the new tunnels.

The larger markets across the rivers lie close to Manhattan. The same MTA paper counts “over 400,000 daily trips” between Manhattan, Queens, and the nearest New Jersey counties. Census data for 2023 count 30,840 private-sector workers who live in New Jersey and work in Brooklyn, 23,645 who work in Queens, and 21,916 who work on Long Island.

Wright made this point himself in January 2014. He told POLITICO that the case for through-running did not rest on riders going from New Brunswick to New Haven. It rested on Penn Station: “Through-running trains would create more capacity.”

New service produces trips that earlier counts could not show. Transport for London and the Department for Transport found in May 2025 that the Elizabeth line is driving about 71,000 additional public transport trips each weekday.

RPA has argued before that a direct ride past Penn Station matters. Its March 2008 report on an earlier Hudson tunnel plan, called ARC, recalls its position that NJ Transit's designs “were insufficient because they did not continue directly to the East Side of Midtown, where large numbers of west-of-the-Hudson commuters were destined.” The report's ridership table puts 30,000 morning riders on an extension to Madison Avenue.

Petra Messick of Amtrak told the August 2024 forum that Gateway's goal is access to Penn Station “with a direct one-seat ride, which is really important to commuters on all lines, all days, all times.”

A through network gives every branch a direct ride to every station on the shared trunk line. The October 2024 study describes how riders reach the other branches: by “disembarking and waiting on the same platform for a subsequent train that runs to their desired destination.”

No published count comes near 100 to 1. The April 2022 Penn Station Master Plan report counts 60,820 passengers getting off and 4,610 getting on at Penn Station in the morning peak hour: 13.2 to 1. NYMTC's 2019 count shows 27,223 rail passengers entering Manhattan from New Jersey against 2,490 leaving in the 8:00 AM hour: 10.9 to 1. Across all rail lines that hour the ratio is 14.0 to 1.

The newest count is lower. NYMTC's 2024 report counts 152,422 rail passengers entering Manhattan from 7 to 10 AM and 18,383 leaving: 8.3 to 1. Counting every way of traveling, it finds that people going against the rush made up “about 26% of morning peak travel.”

The planners' forecast is lower again. The Penn Station Master Plan's June 2020 forecast for the 2042 morning peak hour is 104,400 passengers getting off at Penn Station and 16,900 getting on if Gateway and Penn South are built: 6.2 to 1. If neither is built it is 76,900 and 13,300: 5.8 to 1.

Wright's own figure in May 2026 was “10 people coming in for every one person going out.” Four months later he said close to 100 to 1.

How often trains run depends on the plan. The October 2024 study describes one: “branch lines operating on 15-minute peak headways on each branch, with a total of six branches feeding a common trunk line operating at 24 tph.”

The FRA's longer-term option for Penn Station supports “up to eight regional commuter through-running trains per hour in each direction.” That is a train every seven and a half minutes, in addition to four Amtrak through trains an hour.

Foster Nichols of WSP told the August 2024 forum that trains on a regional metro service “could get to twelve trains per hour per track,” with “a two or three-minute dwell-time.”

RPA's April 2026 report does not compare how often trains would run under a through-running timetable and under today's.

A better concourse adds no trains, and adding trains is Wright's own test. At the August 2024 forum he said RPA comes at Penn Station “from a capacity issue first.” He dropped RPA's call to move Madison Square Garden for the same reason. The federal government would not pay for it, he said, “because it will not add capacity to the system.”

RPA said as much about the last new hall. Its February 2020 paper: “Moynihan is not a panacea for the majority of the problems with Penn Station.”

The project now under way does both. Amtrak's June 2026 agreement with Penn Transformation Partners, the developer, lists open concourses and “the introduction of at least limited through-running on the regional rail network” in one scope of work.

The FRA compares its changes inside the existing station with the plan to expand it. They “could provide up to a third of the additional capacity envisioned under the previous Penn Station Expansion project at a fraction of its estimated $16.7 billion cost.”

RPA's own report says not to wait. “The railroads should start now to address these very real institutional constraints so they are resolved when the tunnels come on line.”

The new tunnel does not add the capacity by itself. Petra Messick of Amtrak told the August 2024 forum that “the tunnel project is more of a resiliency project than a capacity project in itself, if no other investments are made.” The FRA says the new tunnel's capacity “will be underutilized” unless the station's two limits are dealt with.

Through trains have run here with no new construction. RPA's Regional Express Rail plan said through trains between New Jersey and Connecticut “can happen today without any new construction.” RPA's April 2026 report cites the “Train to the Game,” which ran NJ Transit trains on Metro-North's New Haven Line for events at the Meadowlands, and says: “This strategy of revenue-to-revenue through-running should be expanded to include additional sports games and special events.”

Amtrak and its developer are designing the station now. The FRA says the design “will keep this Penn Station Future option open by preserving space for future stairs and escalators and avoiding new structural columns where tracks are removed and platforms are widened.” Amtrak's June 2026 agreement lists “the introduction of at least limited through-running on the regional rail network” among the project's aims.

Wright said the opposite in January 2014. He told POLITICO: “It's that Penn Station is a disaster.” “Through-running trains would create more capacity.” POLITICO reported then that running trains through and turning them at less crowded stations would let 25 percent more trains use Penn Station.

The FRA says through-running adds capacity. “Through-running train operations, both revenue and non-revenue, at Penn Station can materially increase overall throughput by eliminating the time and operational complexity associated with turning trains within the station.” Turning a train around at the station, the report says, “can add 15 to 25 minutes or more under constrained conditions.”

RPA's April 2026 report says the same of the existing station: “Limited applications of revenue-to-revenue through-running could be employed to modestly increase capacity at Penn Station during the shoulder of the peak periods and serve sports and special events at the Meadowlands.”

RPA said so in August 2017. Its report that year faulted Gateway because it “limits service options by maintaining Penn Station as a terminus, rather than allowing through running between New Jersey and Long Island.” It proposed changes to the existing station “to increase station capacity, reduce congestion at the platform level and enable through-running regional rail.” First on its list: “Widen select existing PSNY platforms.” It estimated that Gateway run through to Queens would raise capacity under the Hudson by 138 percent. Gateway ending at Penn Station doubles it.

RPA said so again in February 2020, in a paper that lists Wright among its authors. “Through-running would improve capacity by providing travelers with direct crosstown service between New Jersey and Long Island and reduce the amount of time trains spend boarding at Penn Station.” The same paper credited through-running with “increasing the number of trains and passengers that Penn Station can handle.” Exhibit B sets these statements in order.

Wright states the conclusion and, in the same sentence, names what is missing from it. No one yet knows how much the existing station can handle. The FRA's Phase II is meant to find out. The FRA says Phase II will “evaluate the potential to reach a higher maximum number of trains per hour” and “will be completed in 2028.”

The FRA has not found the limit. “Additional station capacity is feasible; however, exceeding a potential maximum of 32 TPHPD requires broader, regionwide improvements that address the reliability and consistency of trains arriving at the station.” Phase II will study those improvements.

RPA's own report leaves the kind of expansion open. It calls for “a station expansion or some other form of system expansion.” Wright told Crain's New York Business in August 2025 that RPA does not support the southern expansion of the station that Amtrak proposed and “has not endorsed any specific plan.” Until Phase II reports in 2028, no one knows how large the gap is.