Buried in the source notes of the Federal Highway Administration’s work zone statistics page is a line that governs how every figure above it should be read. National metrics do not exist for the number of work zones occurring annually, or for the vehicle-miles of exposure to work zones. The outcomes are counted. The thing the outcomes are outcomes of is not.
That absence explains why the subject is argued with totals rather than rates. The same FHWA page does attempt one exposure figure, and its arithmetic is worth following, because it shows what a rate costs when the denominator has to be improvised. One work zone fatality per 4 billion vehicle-miles of travel, the page reports, and one per $112 million of roadway construction expenditure. The source notes give the inputs: 799 work zone fatalities in 2017, 3,174,408,000,000 vehicle-miles in 2016, and an average estimate of $89.3 billion in road construction spending in 2017. Divide and both ratios come out. They also cross a year boundary, pairing 2017 deaths against 2016 travel.
Who actually dies in a work zone
FARS, the Fatality Analysis Reporting System run by the National Highway Traffic Safety Administration, records a crash as a work zone crash when the first harmful event occurred inside the boundaries of a work zone or on an approach to or exit from one, and resulted from an activity, behavior, or control related to moving traffic through that work zone. The definition is procedural rather than physical. It turns on how a police officer coded a form, and FHWA says as much in its notes on the data: the figures are only as accurate as the crash report forms they come from, and some crashes that did occur in work zones may not have been recorded that way. Whether that undercount runs one direction is not something FHWA’s note settles; the same approach-or-exit language that can miss a work zone crash can also fold in a crash that a stricter physical boundary would exclude, so the direction of the resulting error is not documented.
For 2022, the values FHWA reported as of April 2024 were 891 work zone traffic fatalities in 821 fatal crashes. Of the people killed, 742 were drivers and passengers, 145 were on foot or on bicycles, and 4 were other. Highway worker fatalities came to 94.
Roughly nine in ten people who died in a work zone in 2022 were not working there. The public image of this problem is a flagger beside a lane closure, and the flagger is real, but the dominant casualty is the driver who arrives at a queue too fast. Rear-end collisions accounted for 174 of the 821 fatal crashes, and speeding was a factor in 281 of them, or 34 percent. By roadway type the crashes split 405 on arterials, 321 on Interstates, 63 on collectors and 32 on local roads, which places most of the harm on roads that were never closed to through traffic in the first place.
Positive protection is a regulation, not a preference
Federal rules give the physical shielding of workers a defined name. Under 23 CFR 630 subpart K, positive protection devices are “devices that contain and/or redirect vehicles and meet the crashworthiness evaluation criteria contained in National Cooperative Highway Research Program (NCHRP) Report 350.” Alongside them the rule defines exposure control measures, which are “traffic management strategies to avoid work zone crashes involving workers and motorized traffic by eliminating or reducing traffic through the work zone, or diverting traffic away from the work space,” and a residual category of other traffic control measures covering everything else.
The hierarchy in that trio matters more than the hardware does. Eliminating the traffic is a better answer than surviving it, which is why full closures and detours belong to the same regulatory family as concrete barrier.
On when to install barrier, the rule declines to give a threshold. “The need for longitudinal traffic barrier and other positive protection devices shall be based on an engineering study,” it says, then lists the situations that place workers at increased risk and should prompt consideration: work zones with no escape from traffic, such as tunnels and bridges; durations of two weeks or more; anticipated operating speeds of 45 mph or greater; workers positioned close to open lanes; and roadside drop-offs left overnight or longer. A separate list of fifteen factors runs from traffic volume and vehicle mix through escape paths for workers to the hazards created by placing and removing the device itself.
There is a dating problem in that definition, and it is the sort of thing only a specialist notices. The regulation still points at NCHRP Report 350, the crash-test protocol that the AASHTO Manual for Assessing Safety Hardware superseded. Colorado’s summary of the implementation schedule puts the deadline for temporary work zone devices, including portable barriers, at December 31, 2019. A designer buying barrier today is working to MASH while the rule that requires the barrier describes it by the older standard. The engineering has moved and the citation has not, which is worth knowing before quoting the definition in a specification. The testing regime behind it is taken up in roadside barrier design.
The plan that has to exist before the cones go out
Subpart J of the same part governs the paperwork, and the paperwork is where the traffic decisions actually get made. Every federal-aid project needs a transportation management plan. Projects designated significant need all three of its components: a temporary traffic control plan, keyed to Part 6 of the Manual on Uniform Traffic Control Devices; a transportation operations component covering demand management, corridor and network management and safety enforcement; and a public information component. Projects that are not significant may get by with the traffic control plan alone.
Significance is left to state policy and engineering judgment, measured by whether the sustained work zone impacts exceed what the state considers tolerable. One trigger is absolute rather than judged. Interstate projects inside a designated transportation management area that occupy a location for more than three days with lane closures are significant by rule. That threshold is short enough to catch a great deal of routine maintenance, which is the point.
What speed enforcement has actually demonstrated
Speed cameras in work zones are an FHWA proven safety countermeasure, and New York State’s program is the best-documented American example, because its authorizing legislation from September 2021 obliged the state to report to the Governor on how the program performed. FHWA’s case study on it, published in January 2025, is unusually specific.
Full enforcement began on May 17, 2023, after a thirty-day warning period. Thirty mobile units operate on controlled-access highways, twenty for the state transportation department and ten for the Thruway Authority, and a notice of liability is issued at 11 mph or more over the posted limit. Across 2,319 deployments from April to December 2023 the units logged 200,030 speeding incidents, rejected 48,369 of them, issued 6,120 warnings and 146,077 notices of liability, and recorded a 7.1 percent repeat offender rate.
Two figures from that record deserve quoting rather than summarizing. Nine percent of the notices went to vehicles travelling more than 20 mph over the limit, and one was clocked at 139 mph in a work zone. Meanwhile the share of vehicles exceeding the limit by more than 10 mph in camera-equipped work zones ranged between 3.5 percent in July 2023 and 2.3 percent that December. Both things are true at once. Compliance is high and the tail is extreme, and it is the tail that kills.
The operational constraints matter more to an agency than the violation counts do. Workers have to be present for a camera to operate. A deployment needs the unit stationary for at least two hours to justify the setup, which rules out most short-term maintenance. And the money does not balance neatly: cumulative program expenditure of $6.86 million across 2021 to 2023 ran against $5.4 million collected in 2023 alone, which the case study reports as a cumulative net cost of $2.4 million.
What the case study does not report is a crash effect. It gives speeds, violations, adjudications and finances. The legislation requires crash data for eligible roadways with and without cameras, but the published federal case study stops short of attributing any change in crashes to the cameras, and an agency citing this program as crash-reduction evidence is going beyond what the document supports.
Why trucks show up disproportionately
The one population that is unambiguously overrepresented is commercial vehicles. An action plan prepared for FHWA, FMCSA and NHTSA by the Texas A&M Transportation Institute in 2023 put the comparison plainly. Commercial motor vehicles were involved in 29 percent of fatal work zone crashes in 2016, rising to 33 percent by 2019 and standing at 27 percent in 2020, against a non-work-zone rate that has held at roughly 13 percent. Large trucks, not buses, are 98 percent of that involvement. Of the people killed in these crashes, 71 percent were in something other than the commercial vehicle.
The report offers three mechanisms rather than one. Lane closures, restricted widths and shortened merge areas degrade exactly the geometry a long combination vehicle needs. Truck travel and work zone activity are concentrated on the same primary roadways at overlapping times. And the work zone itself generates heavy truck trips, hauling material in and spoil out, so the facility carries more trucks than it did before the project started.
The crash geometry follows from the physics. The front and back of the commercial vehicle were the initial point of contact in 47 and 38 percent of these crashes respectively, 52 percent occurred on Interstates, and the most common type was a rear-end collision at 46 percent. The same asymmetry appears away from work zones and is examined in large-truck crash data.
The intervention nobody counts as a safety measure
Every treatment described above operates inside a work zone that has already been scheduled. The measure that removes the exposure entirely is a pavement that does not need rebuilding as often, and it is never filed under safety. Postpone a corridor’s first rehabilitation by a decade and one work zone disappears from its history, taking queue, taper and crash record. The engineering behind that is long-life pavement design, and the scheduling question sits in night highway construction, where the trade-off is between fewer daytime queues and worse visibility for everyone present, a trade the safety section weighs alongside every other countermeasure it covers.
Counting that benefit would require the denominator FHWA says does not exist. Until someone builds it, durability will keep being argued as a carbon and cost measure, and its largest single benefit will keep going unpriced.