What the Transit Score measures

The Transit Score is a screening tool. It answers whether an area’s land use could justify studying a given kind of transit investment — never whether that investment should be built.

Transit works where trips concentrate. Concentration comes from three things: people and households packed closely enough that a walkable catchment holds enough riders; jobs packed closely enough that trips have a destination worth serving; and households whose access to a car is limited enough that transit is a first choice rather than a last one. The index measures all four as densities per acre of land, weights them, and reduces them to one number per tract.

Because it is expressed per acre of land, the score is directly comparable between a 40-acre tract in Hoboken and a 12,000-acre tract in the Pinelands. And because the categories are anchored to dwelling-unit densities, a municipality can read a score change directly as a zoning change.

The five categories

Shares are of the statewide total in each column, from 2020–2024 data. Three quarters of New Jersey’s population lives on a fifth of its land at a score of 1.0 or above.

The index and its calibration

Four densities, five parameters, one number. Every parameter is either a measured quantity or a stated planning judgement, and both are set out below.

Household-size divisor — 2.62

The population term expresses residents in household-equivalents, so the divisor is the average household size. ACS table B25010 puts New Jersey’s average at 2.62 persons per occupied unit for 2020–2024. Household size is falling, which matters: a given number of dwellings now houses fewer people than it did a generation ago, so more dwellings are needed to reach the same ridership base.

Zero-car weight — 2.816

Households without a car are weighted above households with one. The weight is measured rather than assumed. ACS table B08141 cross-tabulates means of transportation to work against vehicles available, which gives the observed transit propensity of each group directly.

A New Jersey worker in a car-free household is 2.816 times as likely to commute by transit as one in a single-car household. That ratio is the weight. Statewide, 11.11% of households have no vehicle and 35.07% have exactly one.

Employment coefficient — 1.2

Jobs are weighted at 1.2 relative to households. This is a planning judgement about how employment and housing compare as generators of transit demand, not a measured quantity, and it is stress-tested in Section 4.

Normaliser — 4.0

The normaliser sets the scale, and is chosen so that the category thresholds correspond to meaningful physical densities rather than arbitrary index values. It is solved so that a wholly residential tract with statewide-typical characteristics reaches a score of 1.0 at 1.5 gross dwelling units per acre — the density at which minimum-intensity all-day local bus becomes defensible. The solution is 3.995, rounded to 4.0.

Density anchors. Gross density is dwellings per acre of total land. Net density assumes residential uses occupy 60% of built-up land after streets, non-residential uses and constrained land are accounted for. Densities shown are for a wholly residential tract carrying the statewide average household size and vehicle-availability mix; a tract with employment or with above-average car-free households reaches a given score at a lower residential density.

Investment criteria — Tables 1-1 and 1-2

Table 1-1 screens: find the category, read across, and see which modes are on the table at all. Table 1-2 sets out the conditions each one must then satisfy.

Table 1-1 — Applicability of investments by Transit Score category

Legend. Y applicable with no further land-use condition  ·  C conditional, applicable only if the criteria in Table 1-2 are met  ·  — not warranted on land use alone. Applicability reflects land use alone. Meeting a condition in Table 1-2 establishes that a mode is worth studying; it does not establish feasibility, affordability or benefit.

Table 1-2 — Detailed criteria and conditions

Employment thresholds by municipality

Several criteria in Table 1-2 turn on the number of jobs in a centre or municipality. These are measured from LEHD LODES8 Workplace Area Characteristics for 2023, aggregated from census blocks to municipality.

60,000 or more jobs — 5 municipalities

• Newark city Essex · 158,616 jobs
• Jersey City city Hudson · 134,993 jobs
• Edison township Middlesex · 84,551 jobs
• Trenton city Mercer · 73,874 jobs
• Woodbridge township Middlesex · 63,392 jobs

30,000 to 59,999 jobs — 24 municipalities

• Parsippany-Troy Hills township Morris · 56,292 jobs
• Lakewood township Ocean · 54,136 jobs
• Cherry Hill township Camden · 52,998 jobs
• Elizabeth city Union · 51,512 jobs
• New Brunswick city Middlesex · 50,937 jobs
• Paramus borough Bergen · 44,730 jobs
• Paterson city Passaic · 42,912 jobs
• Mount Laurel township Burlington · 42,448 jobs
• Hackensack city Bergen · 41,968 jobs
• Atlantic City city Atlantic · 41,534 jobs
• Hamilton township Mercer · 41,089 jobs
• Franklin township Somerset · 39,962 jobs
• Toms River township Ocean · 39,273 jobs
• Bridgewater township Somerset · 38,972 jobs
• Camden city Camden · 37,653 jobs
• Secaucus town Hudson · 37,042 jobs
• Clifton city Passaic · 36,280 jobs
• Morristown town Morris · 35,517 jobs
• Wayne township Passaic · 34,418 jobs
• Vineland city Cumberland · 34,106 jobs
• Piscataway township Middlesex · 33,129 jobs
• South Brunswick township Middlesex · 32,334 jobs
• Union township Union · 32,259 jobs
• Evesham township Burlington · 30,735 jobs

Reading the lists. Newark, at 158,616 jobs, is the only New Jersey municipality clearing the 150,000-job threshold the criteria set for heavy rail rapid transit. Atlantic City, at 41,534 jobs, sits in the 30,000 tier. LODES counts jobs covered by state unemployment insurance and excludes most self-employment, uniformed military and some federal employment, so these are conservative counts.

Cost criteria

Cost bands are anchored on current, publicly documented New Jersey capital estimates rather than on escalated historical unit costs.

Costs are as published, in year-of-expenditure dollars. At-grade light rail on an existing freight alignment — the least expensive fixed-guideway product available in New Jersey — currently runs at roughly $89–100 million per route mile. Screening that relies on historical unit costs will systematically overstate what is affordable.

Validation and sensitivity

An index that sorts places by transit potential should predict transit use. The test below uses journey-to-work data that forms no part of the index.

External validation against journey-to-work data

ACS table B08301 reports how workers actually travel to work. It is not an input to the index. If the index measures something real, observed transit commute share should rise across the five categories in order. It does, and steeply.

Observed transit commute share by Transit Score category, New Jersey, 2020–2024. Statewide, 8.12% of 4,582,367 workers commute by public transportation. A worker in a High-score tract is 10 times as likely to use transit as one in a Low-score tract. The relationship is monotonic across every band, which is the property a screening index most needs.

Reading the correlations. The composite index outperforms every one of its own components, though only modestly — 0.629 against 0.62 for population density on its own. The four-factor structure earns its place without dominating. The index is built to be transparent, reproducible from public data and interpretable in units a planning board can act on, rather than to maximise predictive power. Employment density is the weakest single predictor of residential transit commuting, as expected: it measures where trips end, not where the residents being scored live.

Sensitivity to the parameters

Each parameter is moved well beyond any plausible error and the statewide category distribution recomputed.

Percentage of statewide population in each category under alternative parameters. Across the entire tested range the share of New Jerseyans living where the score is 1.0 or higher moves only between 72.7% and 79.5%. The index is not delicately balanced on its parameters; the geography it describes is robust to them.

Who the index describes

The Transit Score is built from land use, not from demography. It contains no measure of income, race or tenure. Yet the populations it sorts are sharply different from one another, and the gradient runs one way.

Population characteristics by Transit Score category. Poverty from ACS table B17001, race and ethnicity from B03002, tenure from B25003, all 2020–2024 five-year estimates. The bar shows each category's share of the state's 389,559 car-free households.

Statewide results

Three quarters of New Jersey’s population and four fifths of its jobs sit on roughly a fifth of its land in areas scoring 1.0 or above. The state’s transit-supportive geography is compact, continuous and concentrated in the north-east.

Shares are of the statewide total in each column and year.

Forecast accuracy, 2000–2024

In October 2000 NJ TRANSIT published a forecast of how New Jersey’s population, employment and land area would distribute across these five categories by 2020. That forecast can now be measured against outcomes. The comparison below holds the measuring instrument constant — the same four factors, weights and thresholds used in the forecast — and changes only the world being measured.

Forecast shares are as published in the 2000 documentation for the year 2020. Actual shares apply the same index and thresholds to ACS 2020–2024 and LODES8 2023 data.

County profile

Ordered by population living in Medium-High or High tracts. Hudson County functions as a single continuous High-score district: 83.4% of its residents live in tracts scoring above 9.0. Sussex and Salem contain no High-score tract.

The fifteen highest-scoring tracts

All fifteen are in Hudson County, which pairs the state’s highest residential densities with its highest rate of car-free households. The index weights both.

Service against potential

The index says what an area's land use could support. Published timetables say what it actually receives. Comparing the two turns a descriptive measure into a diagnostic one.

Method

Every scheduled weekday departure in NJ TRANSIT's published rail and bus GTFS feeds was counted at its stop, and stops were assigned to census tracts by point-in-polygon. A single representative weekday was selected as the date carrying the largest number of active services, giving 21,669 rail departures and 794,786 bus departures across 1,675 tracts. Departures are stop events, so a tract with four routes at hourly headways over a sixteen-hour span records more than one with a single frequent route.

Scheduled service by Transit Score category. The gradient is monotonic and steep: median weekday departures rise from 0 in Low-score tracts to 556 in High-score tracts. This is a second external validation — service planners, working without reference to this index, have allocated service in almost exactly the order it predicts.

Where service falls short of potential

A tract is flagged as underserved where it scores 1.0 or above and receives fewer than 40% of the median weekday departures for its own category. The threshold is a convention, stated so it can be varied; the comparison is internal to each band, so it does not penalise a Medium tract for lacking the service a High tract gets.

468 tracts, home to 1,913,666 people, receive less than two fifths of the service typical for their score band. Seventy-nine of them score above 9.0 — the band in which Table 1-2 calls for all-day service at twenty-minute headways.

Ordered by population. The concentration in Ocean County reflects Lakewood and its surrounding townships, where residential density has risen far faster than the bus network serving it.

Scoring on populated land

The index divides by total land area, which understates tracts containing large tracts of forest, wetland, watershed or military reservation. That effect can be measured. Using 2020 Census blocks, the land in blocks with at least one resident is 83.2% of New Jersey's land area; the median tract is 95.3% populated. Recomputing the index on populated-block land only moves 214 tracts up at least one category.

Ordered by population. The populated-land score is published alongside the primary score in the accompanying dataset. It is a coverage proxy, not a regulatory determination of developability: a block with one resident counts as populated, and zoning, wetlands and preservation designations are not consulted. The primary score remains the one calculated on total land, for comparability.

The 2050 outlook

The outlook is a scenario, presented separately from the observed score so the two are never confused. It applies growth that the three metropolitan planning organisations have formally adopted.

Method

1. Adopted municipal forecasts are collected from all three New Jersey metropolitan planning organisations: NJTPA for its thirteen northern counties (municipal population, households and payroll employment to 2050, September 2025); DVRPC for Burlington, Camden, Gloucester and Mercer (municipal population and households, 2050 v2.1, adopted May 2025, with matching municipal employment); SJTPO for Atlantic, Cape May, Cumberland and Salem (county population and employment to 2050).
2. Growth ratios are computed as forecast 2050 ÷ forecast 2020 for each municipality and variable, then raised to the power 26 ⁄ 30 so that only the 2024–2050 portion of forecast growth applies to the observed 2024 base. Ratios are clipped to 0.3–4.0 so that a small municipality with an unstable base cannot distort a tract.
3. Municipal ratios are allocated to tracts using 2020 Census block populations. Each census block is assigned to both its tract and its municipality, giving a population-weighted split across the 2,235 tract × municipality cells that exist statewide. Fifty-one New Jersey tracts straddle a municipal boundary; for the remainder the allocation is exact.
4. Zero- and one-car household counts grow at the household growth rate, holding each tract’s vehicle-availability mix constant.
5. The index is recomputed on the 2050 values with identical parameters.

Households grow faster than population (+12.0% against +9.4%) because average household size continues to fall in every adopted forecast — a structural change that raises transit potential per resident, since the index counts households as well as people.

Applying the score to a land-use decision

The score’s most practical use is testing a zoning change before it is adopted. Because the categories are anchored to dwelling-unit density, a proposed change can be scored directly.

The situation. A 1,000-acre developing area is zoned for 1,000 housing units. The municipality wants to know whether it can justify better bus service, or eventually light rail, to an older employment centre nearby with 40,000 jobs.

Result A. A score of 0.58 is Marginal. Table 1-2 permits peak-period-only extensions of existing bus routes and surface park-and-ride for express bus or commuter rail. It does not support all-day local bus, and it does not support light rail in any form.

Result B. Tripling the housing yield and adding a small amount of ground-floor employment moves the area from Marginal to Medium. That unlocks minimum-intensity all-day local bus — an 8 to 12 hour span at 30 to 60 minute headways — plus local circulator service and, if the area ends up surrounded by Medium-High tracts, conditional consideration of medium/low-cost light rail. It does not on its own justify light rail to the 40,000-job centre; that requires the corridor to reach Medium-High, roughly 4.51 gross units per acre.

Rail investment in New Jersey, 2000–2026

Thirty-four fixed-guideway and priority-transit projects were identified for New Jersey in October 2000, across three tiers: committed and funded, candidate, and for definition or study. Each has been traced to its present disposition against NJ TRANSIT's current project inventory and the NJTPA FY 2026–2029 Transportation Improvement Program.

Funding at the point of listing predicted delivery almost perfectly. Of the nine projects already committed and funded in 2000, eight were completed and one — the Hudson–Bergen Light Rail — was delivered in part, its core system built and its northward expansion still outstanding. Of the nineteen candidate projects, two were completed, one delivered in part, five remain active in delivery or study, ten are dormant, and one was cancelled and superseded. Of the six listed only for definition or study, none advanced. A Transit Score establishing that a corridor merits study remains necessary; it has never been sufficient.

Corridors drawn on the interactive map

Ten of the thirty-four proposals would have run along a rail right-of-way that still exists and is still named. For those, the interactive map draws the corridor itself rather than a marker, using the alignment recorded in OpenStreetMap under the Open Database License. The extent shown reflects how completely each corridor is tagged there, so a few are drawn shorter than the route a project would have used. No vector alignment was published for the 2000 map, so no attempt is made to reproduce the lines as they were drawn then.

Disposition of each project

Traced as of August 2026. “Dormant” indicates that no current sponsored phase was located in the reviewed NJ TRANSIT inventory or the FY 2026–2029 TIP; it does not imply formal cancellation. “Not advanced” indicates the concept as scoped in 2000 was not carried forward, even where related investment has since occurred nearby.

Current pipeline

Thirteen NJ TRANSIT expansion and capacity projects are in delivery, design or study: 5 under construction, 2 in near-term design and 6 identified for the longer horizon. Scope and cost figures are drawn from NJ TRANSIT’s published capital plan project sheets and from federal and sponsor documentation.

Status current as of August 2026 and subject to change. Cost figures are stated in the dollars of the year each estimate was published and are not adjusted to a common basis. Inclusion is a statement of public record, not an endorsement, and does not imply that a project satisfies the criteria in Table 1-2.

Limitations and sources

Limitations

• Land area is not developable area. The index divides by total land area, so tracts containing large protected or undevelopable holdings — the Pinelands, Wharton State Forest, the Meadowlands, military reservations, major watershed land — are scored as though that land were available for development. The effect is quantified in Service against potential, and a populated-land score is published alongside the primary score for every tract. It remains a coverage proxy rather than a determination of developability.
• Service counts are stop events on one representative weekday. They do not distinguish a frequent route from several infrequent ones, weight by span of service, capacity or reliability, or account for a stop serving residents of an adjacent tract. Seasonal, school and shuttle services scheduled outside the selected weekday are excluded.
• Demographic characteristics are reported, not modelled. Section 5 describes who lives in each category; it does not establish that transit investment caused, or would alter, those patterns.
• LODES employment is a partial count. It covers jobs reported through state unemployment insurance and excludes most self-employment, uniformed military and some federal employment. Employment density, and therefore the score, is conservative where the employment mix is unusual.
• ACS estimates carry sampling error. Five-year tract-level estimates have margins of error that can be wide for small counts, particularly zero-vehicle households in low-population tracts. Margins are published in the source tables and are not propagated through the index. Individual tract scores close to a threshold should not be treated as precise.
• The 2050 outlook holds spatial distribution constant within municipalities. It reflects aggregate adopted growth, not redevelopment geography.
• Vehicle-availability shares are held constant to 2050. No adopted forecast publishes tract-level vehicle availability, so each tract retains its 2024 mix. If car ownership continues to fall in dense areas, the outlook understates 2050 scores there.
• The index measures land-use potential only. It contains no measure of existing service, travel time, right-of-way availability, cost or willingness to pay.
• Six water-only tracts with zero recorded land area are excluded, leaving 2,175 of New Jersey’s 2,181 tracts.

The index in full

Sources

• PopulationU.S. Census Bureau, American Community Survey 2020–2024 5-Year Estimates, table B01003, via the ACS Summary File.
• Households and vehicle availabilityACS 2020–2024 5-Year, table B25044 (tenure by vehicles available). Zero-vehicle households are B25044_003 + B25044_010; one-vehicle households are B25044_004 + B25044_011.
• Average household sizeACS 2020–2024 5-Year, table B25010, New Jersey: 2.62 persons per occupied unit.
• Zero-car weightACS 2020–2024 5-Year, table B08141, means of transportation to work by vehicles available, New Jersey.
• Scheduled serviceNJ TRANSIT published rail and bus GTFS feeds, counted as weekday departures per stop on the representative weekday carrying the largest number of active services, and assigned to tracts by point-in-polygon.
• Poverty, race and tenureACS 2020–2024 5-Year, tables B17001 (poverty status), B03002 (Hispanic or Latino origin by race) and B25003 (tenure).
• Populated land2020 Census P.L. 94-171 block population joined to TIGER/Line 2020 block land area; populated land is the area of blocks with at least one resident.
• Validation dataACS 2020–2024 5-Year, table B08301, means of transportation to work. Public transportation is B08301_010; workers 16 and over is B08301_001.
• EmploymentU.S. Census Bureau, LEHD LODES8 Workplace Area Characteristics, New Jersey, 2023, aggregated from 2020 census blocks to tract and municipality using the LODES geography crosswalk.
• Land area and boundariesU.S. Census Bureau TIGER/Line 2024: census tracts (ALAND), counties, county subdivisions, primary and secondary roads, rails and area water.
• Tract to municipality allocation2020 Census P.L. 94-171 Redistricting Data, block-level total population, joined to municipalities through the LODES geography crosswalk. 2,235 tract-by-municipality cells; 51 tracts straddle a municipal boundary.
• Rail network and stationsNJ TRANSIT published rail GTFS feed: 231 stations and the shape geometry of all 17 rail and light rail services.
• 2050 forecasts, northNorth Jersey Transportation Planning Authority, county and municipal population, household and payroll employment forecasts to 2050, September 2025.
• 2050 forecasts, south-westDelaware Valley Regional Planning Commission, Adopted 2050 v2.1 Population & Employment Forecasts, adopted 22 May 2025, municipal files for Burlington, Camden, Gloucester and Mercer counties.
• 2050 forecasts, south-eastSouth Jersey Transportation Planning Organization, Moving South Jersey Forward RTP 2050, Appendix C, for Atlantic, Cape May, Cumberland and Salem counties.
• Hudson Tunnel Project costFederal Transit Administration, Capital Investment Grants quarterly monitoring report, NJ–NY Hudson Tunnel Project, June 2025: total project cost $16.041 billion.
• Portal North Bridge costFederal Transit Administration Capital Investment Grants project profile, Portal North Bridge, Hudson County: $1.787 billion in year-of-expenditure dollars.
• NJ TRANSIT capital programmeNJ TRANSIT, Capital Plan project sheets, for project scope and cost estimates; NJ TRANSIT project inventory and reports.
• Programmed fundingNJTPA, FY 2026–2029 Transportation Improvement Program detail reports, approved 8 September 2025, used to establish whether a project has a current sponsored phase.
• Glassboro–Camden LineProject sponsor materials and contemporaneous reporting: 18 miles, estimated capital cost $1.6–1.8 billion.
• Method lineageNJ TRANSIT Business Planning Department, The 2020 Transit Report: Possibilities for the Future, October 2000, and the accompanying 2020 Transit Map. Primary author of the Transit Score Index: Thomas Marchwinski.
• Underlying researchJeffrey Zupan, Public Transportation and Land Use Policy and Urban Rail in America; Delaware Valley Regional Planning Commission, Transit Potential in the New Jersey Counties (1993); NCHRP Report 365, Travel Estimation Techniques for Urban Planning; Richard Voith, Federal Reserve Bank of Philadelphia, on auto ownership near rail stations.