Traffic congestion happens when more vehicles try to use a road than it can handle at a steady speed. That gap between demand and capacity is the root cause, but it takes different forms depending on where you are and what time of day it is: bottlenecks baked into the road’s design, crashes that pull lanes out of service without warning, construction that goes on for months, rain or ice that slows everyone down, signals running on outdated timing plans, and stadium events that send thousands of cars into the same corridor at once.
KEY TAKEAWAYS
- Congestion splits into two categories: recurring (predictable bottlenecks and poorly timed signals, ~45% of total delay) and non-recurring (crashes, weather, work zones, events, ~55%), per FHWA estimates.
- Physical bottlenecks account for about 40% of all congestion nationally — more than any other single cause.
- Traffic incidents cause roughly 25% of national congestion. Each minute a freeway lane stays blocked generates about four minutes of delay after it reopens, per FHWA’s Traffic Incident Management program.
- US drivers lost 49 hours to traffic in 2025, up from 43 hours in 2024, and congestion grew in 254 of the 290 cities INRIX analyzed.
- Congestion costs the US trucking industry approximately $109 billion annually in lost productivity and fuel, according to the American Trucking Associations, citing ATRI data.
American drivers felt all of it in 2025. The average US driver lost 49 hours to congestion last year, up six hours from 2024, costing roughly $894 per driver and $85.8 billion nationally in lost time, according to INRIX’s 2025 Global Traffic Scorecard. Congestion grew in 88% of the 290 US cities INRIX tracked, and Chicago overtook New York as the most congested city in the country, with drivers there losing 112 hours and about $2,063 each.
In this guide, we cover the main causes of traffic, why recurring and non-recurring congestion need different responses, how incidents create delays that outlast the incident itself, what 2025 data shows about US congestion, and how commercial fleets reduce their exposure to it.
Recurring vs. non-recurring congestion
Transportation researchers split congestion into two buckets. The split matters because fixing one type does nothing for the other.
Recurring congestion shows up at the same place and the same time every single day. A freeway on-ramp that was built when the metro area was half its current size. An intersection where the signal timing was programmed in 2009 and has never been touched since. Commuters who drive these roads every morning learn to expect it and plan around it. That predictability is cold comfort, but it does make it manageable.
Non-recurring congestion has no schedule. A crash at mile marker 47. A water main break that takes three lanes out of service at 7 AM. A storm that drops two inches of rain on a city with no drainage capacity. Drivers cannot plan for any of it, which is why this category tends to feel worse even when the raw delay time is shorter. Unpredictability carries its own cost.
FHWA puts recurring congestion at about 45% of total national delay and non-recurring at about 55%, drawing on research from the Texas Transportation Institute and other transportation groups.
Quick reference: Causes of US traffic congestion
| Cause | Share of national delay | Type |
| Physical bottlenecks | ~40% | Recurring |
| Traffic incidents | ~25% | Non-recurring |
| Weather | ~15% | Non-recurring |
| Work zones | ~10% | Non-recurring |
| Poor signal timing | ~5% | Recurring |
| Special events | ~5% | Non-recurring |
Physical bottlenecks
A bottleneck is any point where a road’s capacity falls below what drivers are demanding from it. A three-lane highway merges to two. An on-ramp dumps cars into a stretch that is already running at the edge of what it can handle. A bridge built to a lower posted speed. Any corridor designed for a city a quarter the size of what it now serves.
Bottlenecks account for about 40% of all national congestion, making them the single largest cause in the country per FHWA data. They are also the most permanent problem on the list. A crash clears. A storm moves through. A bottleneck is in the same location the following morning, and the morning after that, until someone spends the money and time to redesign the road or until enough demand shifts elsewhere on its own.
Chicago illustrates what geography does to bottleneck formation. Lake Michigan cuts off eastward expansion entirely, so a metro area of nearly 10 million people funnels through a limited set of north-south and east-west corridors. That physical reality, not just infrastructure quality, is a large part of why Chicago posted the worst congestion numbers in the country in 2025.
The numbers for commercial trucks are worse than general driver data suggests. ATRI’s 2026 Top 100 Truck Bottlenecks report confirms this specifically: the I-294 and I-290/I-88 interchange west of Chicago is now ranked the single most congested point in the country for commercial truck traffic, overtaking the George Washington Bridge corridor in New Jersey for the first time in the report’s 15-year history. The interchange handles more than 300,000 vehicles a day and sits in year six of an $800 million reconstruction project scheduled for completion in 2027. ATRI estimates that congestion at locations like this is the equivalent of 436,000 truck drivers sitting idle for an entire year.
Traffic incidents
Crashes, stalled vehicles, and debris on the road cause about 25% of national congestion, per the US Department of Transportation. Stalled vehicles in particular are often preventable. A truck that fails a pre-trip inspection and breaks down in a travel lane does not just affect that fleet’s delivery window — it creates congestion for everyone behind it.
The delay a blocked lane generates runs well past the incident itself. According to FHWA’s Traffic Incident Management program, each minute a freeway lane stays blocked during peak use produces roughly four minutes of congestion after the road reopens. The queue that was built while the lane was unavailable does not dissolve the moment traffic has room to move. It drains slowly, and on a corridor already running near capacity, that drainage process can stretch an hour past clearance.
USDOT data shows that the best-performing state transportation agencies clear incidents within about 22 minutes of arriving on scene. On a busy corridor, 22 minutes of lane blockage during peak hours can still create congestion that runs well over an hour.
Rubbernecking adds time on top of the physical delay. A vehicle moved completely off the roadway still draws down speeds for drivers passing the scene, and that slowdown ripples backward through traffic for longer than the actual lane closure did.
Work zones
Construction and maintenance projects reduce lane capacity for weeks or months at a stretch. FHWA attributes about 10% of national congestion to work zones, with the worst effects on corridors that were already running near their capacity before the cones went up.
The problem is not limited to delay. FHWA data shows an average of roughly two fatalities and around 101 injuries in US work zones every day. That safety cost is part of why several states have moved to automated speed enforcement on high-volume construction projects, and why work zone speed limits have become a harder enforcement priority for state DOTs.
For fleets running time-sensitive deliveries, work zones are a particular problem because they tend to last longer than the original project timeline and can shift location or lane configuration without much advance public notice. A route that was clean last Tuesday may have lost a lane by Thursday morning.
Weather
Rain, snow, ice, and fog do not physically remove lanes from service. They reduce the effective capacity of the lanes that remain. Drivers slow down, leave more following distance, and brake earlier, which compresses how many vehicles a given stretch of road can process in an hour. FHWA estimates that weather accounts for about 15% of national congestion, third behind bottlenecks and incidents.
Rain and ice also raise crash probability. When a weather delay triggers a secondary incident on the same corridor, the two congestion categories stack on top of each other. One ice storm can simultaneously create weather-related slowdowns and incident-related delays across a wide area at once.
For fleet dispatchers relying on real-time location data from their vehicles, a truck that starts falling behind its planned route during a storm is visible early enough to reroute or, at a minimum, notify the customer before the delivery window is already gone.
Poor signal timing and special events
Signal timing accounts for about 5% of national congestion, the smallest share on this list. It is also the cheapest problem to fix compared to any of the others. Signals calibrated to traffic volumes from years ago, or not coordinated with the intersections they connect to, produce stop-and-go patterns that would not exist with updated programming. Cities that have retimed their networks have documented real reductions in corridor travel time without building anything.
Special events are also roughly 5% of national congestion on average, but the effect is completely concentrated. A single sold-out game can send 30,000 or 40,000 cars into the same set of roads within 90 minutes of the final score. Cities hosting World Cup matches in 2026 are treating this as a logistics challenge equivalent to months of recurring bottleneck delay compressed into a single afternoon, because that is roughly what the data says it is.
What the 2025 numbers show
The surge in 2025 congestion is not primarily a new demand story. It is a return story. Car commuting is nearly back to 2019 levels nationally, per INRIX. Public transit ridership reached 85% of pre-pandemic levels in 2025, per APTA’s April 2026 Ridership Update, meaning roughly 15% of pre-pandemic riders are still not returning to trains and buses. That gap represents a meaningful volume of trips now running on roads not designed to carry them.
The five most congested US metro areas in 2025, per INRIX: Chicago at 112 hours lost per driver, New York at 102, Philadelphia at 101, Los Angeles at 87, and Boston at 83.
New York held flat while nearly every other major city got worse. INRIX analysts point to Manhattan’s congestion pricing program, which launched in 2025, as a likely factor. The Congressional Budget Office projects the Highway Trust Fund’s expenditures will exceed revenue by about $40 billion in FY2029, which means the structural capacity gap is not closing on its own timeline.
What this means for commercial fleets
A passenger car driver stuck in traffic loses time. A commercial fleet stuck in the same traffic loses time, fuel, and sometimes the delivery window itself, all at once.
The scale of this cost is measurable specifically for the freight industry. Congestion costs the US trucking industry approximately $109 billion annually in lost productivity and fuel, according to the American Trucking Associations, citing ATRI data. That makes congestion one of the largest single cost items in commercial fleet operations, ahead of most factors a fleet manager can actually control.
A truck idling in stop-and-go traffic burns fuel without moving freight, a cost that compounds across every delayed route. Fleet fuel monitoring turns that idle time into a measurable number rather than a line item buried in monthly totals. If cost tracking shows the same route consistently losing 20 minutes to a known bottleneck three times a week, that is a measurable fuel and labor cost that route planning can address directly.
A driver caught in an unplanned crash delay may hit hours-of-service limits before finishing a route that looked completely achievable at dispatch time. Managing that exposure is part of what a structured fleet safety program addresses before it becomes a CSA violation.
Delivery consistency is what shippers actually track. A carrier that holds a three-day window reliably gets repeat freight. One running the same average with wide variation around it creates planning problems for its customers, and customers eventually route around that problem by finding someone more predictable. Congestion does not affect every carrier the same way — fleet telematics data is what separates fleets that can document and address their exposure from those that absorb it as a cost of doing business. Part of that picture is driver behavior; how a driver responds to slow traffic affects fuel burn, vehicle wear, and safety in ways that show up in the data long before they show up in an incident report.
How Intangles helps fleets manage around congestion
The dispatcher working a live board cannot manage what they cannot see. A call from a driver saying they are stuck somewhere on the interstate is not the same information as a platform showing the vehicle has been stationary for 19 minutes in a section of I-90 where this happens on Thursday afternoons, with an alternate route flagged. Intangles is a digital twin platform that gives fleet managers continuous GPS position alongside live vehicle diagnostics, so a routing decision and a maintenance flag can happen in the same moment rather than two days apart.
| Capability | What it does |
| Real-time GPS tracking (InRoute) | Shows exactly where a vehicle has stopped and for how long, so dispatchers can tell routine congestion from a developing incident and reroute accordingly |
| Live diagnostics via InGenious (OBD-port install, no vehicle modification) | Flags rising engine temperature or fault codes during extended idling in traffic, before a stall becomes a lane-blocking breakdown |
| Predictive vehicle health monitoring | AI-driven analysis (96% predictive accuracy) flags components likely to fail under stop-and-go load, contributing to up to a 75% reduction in powertrain breakdowns |
| Driver behavior scoring (DriveIQ) | Tracks 20+ behavioral exceptions, including harsh braking in stop-and-go traffic, so coaching addresses the driving patterns congestion creates |
For carriers running urban routes on tight windows, having real-time position and predictive vehicle health monitoring in a single platform means congestion exposure and vehicle reliability get managed together, instead of living in separate systems that don’t share data.
At Intangles, the fleets seeing the clearest insurance impact from the platform are those that bring telematics data into the renewal conversation proactively. A CSA score tells an underwriter what citations a fleet received. DriveIQ data tells them how every driver behaves between citations. That distinction is what moves a fleet from industry-average pricing to a rate that reflects its actual risk profile.
Explore the platform or get in touch with our team to learn how Intangles helps US fleets plan around congestion instead of just reacting to it.
KNOW MORE
Frequently Asked Questions
What is the single biggest cause of traffic in the United States?
Physical bottlenecks account for roughly 40% of all congestion nationally, per FHWA. These are permanent capacity constraints — merges, lane drops, and interchange designs that predate the traffic volumes they now carry. Unlike crashes or storms, they do not clear on their own.
Do accidents cause more traffic than construction?
Yes. Traffic incidents cause about 25% of national congestion, versus about 10% from work zones. The multiplier effect makes incidents particularly costly: each minute a freeway lane stays blocked generates roughly four minutes of delay after it clears, per FHWA.
How much did traffic congestion cost US drivers?
The average US driver lost 49 hours and roughly $894 in 2025. Nationally, that adds up to about $85.8 billion in lost time, per INRIX’s 2025 Global Traffic Scorecard.
Which US city currently has the worst traffic?
Based on the most recent data available, Chicago, where drivers lost 112 hours and about $2,063 each in 2025. New York held flat at 102 hours that year, which INRIX ties in part to Manhattan’s congestion pricing program that launched in 2025.
Can better signal timing actually reduce congestion?
Yes, and it is one of the cheaper interventions available. Poor signal timing accounts for about 5% of national congestion. Updating signal programs to reflect current traffic volumes and coordinating adjacent intersections cuts stop-and-go delay with no construction required.
Why does non-recurring congestion feel worse than recurring congestion?
Because there is no way to plan for it. A driver who knows the same stretch backs up every weekday at 5 PM leaves earlier. A driver who hits a crash scene that did not exist 45 minutes ago cannot do anything except wait. The unpredictability is a separate cost on top of whatever the delay itself takes.
We’re looking forward to meeting you