Two Subways

Same fare. Same map. Two systems.

The MTA's “Customer Journey-Focused Metrics” dataset measures the only question riders actually care about: what share of you, at rush hour, arrived more than five minutes late? Over the twelve months ending April 2026, the answer for the B was 21.4%. The Q was 21.2%. The D, 19.3%. The M, 19.2%.

For the L, it was 9.1%. For the 7, 9.0%.

The worst lines have lagged the system for a decade

The five worst lines for delayed peak journeys - B, Q, D, M, G - all run through the older BMT division, the one whose signal infrastructure was built when Coolidge was president. The best-performing lines - L, 7, JZ - either have modern CBTC signaling installed (L, 7) or run on lighter, shorter routes (JZ).

This isn't anomaly. It's a ten-year trend.

The L's outperformance shows up the moment full CBTC signaling came online. The 7 followed. Every other major line is still running on relay-based signals from the 1930s.

Aging steel is most of the story

When the MTA classifies the cause of each delayed train, one category swallows almost everything else.

“Infrastructure & Equipment” - track, signals, power, rolling stock - is roughly half of every line's delay ledger. “Planned ROW Work” - i.e., announced track shutdowns - is another big slice. Crew availability is the smallest column on every bar.

The story New Yorkers tell themselves about delays (operator error, labor disputes, the homeless) is at most a third of the actual ledger. The dominant cause is the same one the MTA's last four capital plans have asked Albany to fund.

Where the riders actually are

The borough story is where the data gets unfair.

Six years after the pandemic, every borough has clawed its way back to something like 72-75% of 2019 ridership. Except the Bronx, which is stuck at 58.6%.

That's not a small gap. The Bronx in 2024 lost 1.4 percentage points of its share of the entire subway system - a borough that was already only 8% of ridership is now under 7%. 57 million fewer annual trips than 2019, or roughly 229,000 missing weekday trips a day. Nobody else absorbed them. They just stopped.

Manhattan rides home. The Bronx rides to work.

The hourly ridership data tells you who the system is for.

Manhattan ridership peaks at 5pm - the homebound commute, for people who work in Manhattan, not who live there. Brooklyn, Queens and the Bronx all peak in the morning, well before 9am. Bronx ridership peaks at 7am sharp - the shift-worker hour.

When the B and the Q underperform on the morning peak, the riders who absorb it are the ones leaving the Bronx and outer Brooklyn at 6am to make a 7am shift in a Midtown lobby or a hospital basement.

Two out of three stations have no elevator

Of the city's 445 station complexes, only 121 are fully ADA-accessible - 27%. The rest you climb.

The geographic distribution is almost spite. Manhattan, the borough with the deepest pockets and the densest stations, is also the most accessible at 31%. Brooklyn - the second-largest ridership borough - is the worst at 22%. Of Brooklyn's 156 stations, 119 have no elevator.

A wheelchair user trying to cross Brooklyn doesn't ride the subway. They take Access-A-Ride, where the average wait is over an hour and on-time performance is unmeasured.

And it isn't just the small stations. The fifteen busiest stations in the entire system without full elevator access carried 140 million riders in 2024 - combined, that's more than every Bronx station put together (81M).

Union Square alone moved 22.9 million people - more than the four busiest Bronx stations combined. None of them can use the elevator at Union Square, because there isn't one.