MTA Daily Ridership Data: Beginning 2020

The pulse of the largest transit system in the United States, taken once a day since the day the world stopped commuting. MTA Daily Ridership Data: Beginning 2020 is the Metropolitan Transportation Authority's daily ridership and traffic estimate series, published through New York State Open Data (MTA) and catalogued here under Passenger Ground Transportation. Every row pairs a travel date with one of nine mode counters - Subway, Bus, LIRR, MNR, SIR, AAR, BT and, since January 2025, the two congestion-pricing counters, CBD Entries and CRZ Entries.

The series opens on March 1, 2020, which makes it the rare official dataset whose first observations ARE the disaster: the COVID collapse, the halting climb-back and the congestion-pricing era all sit in one table of roughly 17,700 rows, observed through August 19, 2026. In Datadory's catalog of 1,744 datasets across 159 viable industries it scores 9/10 for quality - comfortably above the 7.81 catalog average - with a fully verified field dictionary behind it, and it anchors a fourteen-dataset passenger ground transportation slice averaging 9.07. Get a sample of this dataset cut to the modes and dates you actually model.

Four rows pulled during the August 2026 research pass, all from one travel date:

date : 2026-08-19   mode : Bus    count : 1288294   # systemwide bus boardings
date : 2026-08-19   mode : LIRR   count : 282448    # Long Island Rail Road rides
date : 2026-08-19   mode : MNR    count : 232931    # Metro-North Railroad rides
date : 2026-08-19   mode : AAR    count : 49535     # Access-A-Ride scheduled paratransit trips

Read them as one day, not four numbers: a single Wednesday carries more than 1.28 million bus boardings, and the two commuter railroads together add another 515,379 rides, putting the four sampled modes alone at 1,853,208 trips. The grain is one row per mode per date, so a month of nine-mode history is a few hundred rows and the whole series is about 17,700 - small enough to hold in memory, structured enough to join on date against weather, events or your own sales calendar. The remaining five counters, including the two congestion-zone lines, arrive in the same shape.

Three fields define every row, all marked verified against the publisher's documentation during the August 2026 review:

• date - the travel date the row describes, and the natural join key for calendars, weather and event tables.
• mode - which of the nine counters the row measures: Subway, Bus, LIRR, MNR (Metro-North), SIR (Staten Island Railway), AAR (Access-A-Ride), BT (Bridges and Tunnels), CBD Entries or CRZ Entries.
• count - the daily total estimated ridership or traffic for that mode on that date.

The dictionary is deliberately boring, which is the compliment: nothing to parse, nothing to reverse-engineer, nothing that drifts between extracts. The one casualty of the move to this tidy layout was the retired edition's percent-of-comparable-pre-pandemic-day companion, which rode beside every mode column in the old wide table. Those comparison columns fold under additional fields on request - name them with your sample and they ship alongside the core three.

• Geography: the Metropolitan Transportation Authority service region - New York City plus the Hudson Valley and Connecticut commuter territory served by LIRR and Metro-North. One region, measured exhaustively, down to the paratransit van and the tunnel lane.
• Temporal: March 1, 2020 to present, roughly 2,363 daily observations, observed through August 19, 2026. The two congestion-era counters enter partway through, in January 2025, so deep history is eight modes wide and the recent window is nine.
• Granularity: daily, one row per mode per date. Systemwide totals, not station- or route-level breakdowns - the trade that keeps a 236-million-person metro region legible in a table you can email.

Against the catalog at large - where 62.8% of 1,744 records score 8 or higher and the average sits at 7.81 - this record's 9/10 lands in the top band, backed by verified field definitions and sample rows on the page. See where else it ranks in best passenger ground transportation datasets.

API, files, or your warehouse. Daily, weekly, or hourly.

You pick the channel and the cadence; the three-field dictionary above travels unchanged through all of them. Analysts usually take the whole nine-mode panel as a file extract, because roughly 17,700 rows fits in a notebook's memory let alone a warehouse. Platform teams sync it beside their own sales, staffing or footfall tables so ridership becomes just another joined column keyed on date. Dashboards poll whichever rhythm the question needs. Name the modes, the date ranges and the cadence when you request the sample - the sample ships first either way, and changing pace afterward is a settings conversation, not a re-integration project.

Nine counters, one region, daily grain - a combination that shows up in unrelated teams' workflows:

1. Recovery tracking by mode. Daily curves from March 2020 onward show subway, bus and the two commuter railroads recovering on different clocks; no monthly aggregate catches the divergence.
2. Congestion-pricing evaluation. CBD Entries and CRZ Entries since January 2025, with BT vehicle counts running alongside as the driving counterpoint - a before-and-after design built into the table rather than assembled across sources.
3. Commuter-rail demand modelling. LIRR at 282,448 and Metro-North at 232,931 rides on the sampled day, joined on date to service calendars and fare changes.
4. Footfall proxies for retail and media planning. Ridership levels stand in for NYC activity when timing store staffing, pickups or out-of-home spend.
5. Paratransit planning. The AAR series tracks scheduled Access-A-Ride trips at a 49,535-trip day's scale - the demand signal for an aging city's door-to-door network.
6. Feature engineering. A three-column tidy panel with zero parsing risk drops straight into a warehouse as exogenous variables for any NYC-facing forecast.

Data scientists and ML engineers (relevance 3) get a clean daily panel for recovery-curve and elasticity models - see data scientists use cases. Market researchers and consultants (3) quantify NYC transit demand by mode with estimates keyed to exact dates - market researchers use cases. Developers and data-product builders (3) get a three-column schema that drops straight into products - developers builders use cases. Journalists, academics and students (3) cite dated official counts for pandemic-recovery and congestion-pricing reporting. Investors and quant researchers (2) read Congestion Relief Zone entries as a fast gauge of Manhattan activity - investors quants use cases. E-commerce operators (2) proxy footfall; sales and growth teams (1) time field marketing; competitive intelligence teams (1) contextualize mobility competitors - competitive intel product teams use cases.

Notes that pair well with this page:

• Passenger ground transportation data hub - the pooled view of the slice, fourteen primary datasets from commute mode shares to ride-hail ledgers.
• National Transit Database (NTD) Monthly & Annual Ridership Datasets - roughly 370,000 agency-by-mode monthly rows since 2002: national breadth where this set offers New York depth.
• CTA Ridership - Daily Boarding Totals - the closest structural twin: 9,312 daily Chicago rows split bus versus rail back to 2001.
• NYC TLC Trip Record Data (Yellow/Green Taxi & FHV/HVFHV Uber/Lyft) - the street-level complement: trip-level taxi and for-hire records for the same city this set measures systemwide.
• Mobility Database - Global GTFS & GTFS-Realtime Feed Catalog - 6,487 schedule and realtime feeds when the question widens from counts to networks.
• vs BTS Open Data Portal - Passenger Travel Collection - the head-to-head on depth against breadth.
• The New York State Open Data (MTA) source profile - who publishes the series and what sits beside it.
• Unlinked Passenger Trips (UPT) - the counting vocabulary transit agencies report against, unpacked.
• Best passenger ground transportation datasets - where this record ranks within the slice and what beats it for other jobs.