Almost every argument about American highways ends up resting on a document somebody else wrote. A condition statistic comes out of a federal file with a submission deadline attached to it. A lane width comes out of a book published by a membership association and made binding by a paragraph of regulation. A claim about what a specific project did comes out of a drawing that ranks below the bid documents it was issued with. The three are different kinds of evidence, and treating them as interchangeable is the most common failure in transportation writing.
This section is the reference spine for the rest of the publication. Its job is to describe what each source can actually support, and to say plainly where each one stops.
Four institutions, four kinds of authority
The federal government publishes the data, and it does so from separate offices with separate habits. FHWA’s Office of Highway Policy Information runs the road inventory and the annual statistical series that has appeared every year since 1945. Its Office of Transportation Policy Studies runs the construction cost index. Its Office of Bridges and Structures runs the bridge inventory, which is mid-migration to a new data standard. NHTSA runs the fatality census. Freight is a partnership with the Bureau of Transportation Statistics rather than a purely FHWA product. None of those datasets was built to answer the question most people ask of it, which is why the working guide to the federal datasets is organized around what each one measures and what its own documentation admits.
AASHTO writes the technical standards, and it is an association of state transportation departments rather than a regulator. Its manuals acquire legal force only where a federal regulation names one by edition and incorporates it by reference, which is a narrower and more datable event than the profession’s shorthand suggests. The mechanics of that transfer, and the very different route by which the traffic control manual became a genuine national standard, are set out in the guide to who writes the rules.
The Transportation Research Board occupies a third position, and its cooperative research programs supply the studies the other two cite. The governance is worth knowing before those studies are read as independent. The National Cooperative Highway Research Program was initiated in 1962 by the chief executive officers of AASHTO, is funded by participating member states, and receives technical support from FHWA. Its annual program is built from needs identified by state department staff, by AASHTO committees and by FHWA. Topics of the highest merit are selected by AASHTO’s Special Committee on Research and Innovation, whose recommendations go each year to the AASHTO Board of Directors, the National Academies and FHWA. TRB was asked to administer the work because of its recognized objectivity. The research underpinning federal criteria is therefore commissioned, in large part, by the agencies those criteria bind.
State departments of transportation are the fourth layer and the one that decides most outcomes. Federal design standards reach the National Highway System. Everything else is built to state criteria, and even on federal routes it is a state manual that names who signs when a criterion is not met.
Where the seams are
Three of the live problems in this landscape are structural rather than accidental.
The first is publication lag, which is rarely stated next to the number it applies to. Calendar year 2024 travel by functional system reached the public in a table dated October 2025. A quarterly movement in highway construction prices reached it more than seven months after the quarter closed. Both were current the day they appeared and neither is current now, and nothing on the face of either says so.
The second is access. A regulation can make a copyrighted manual binding without making it readable. Design criteria incorporated by reference can be inspected at two federal locations in Washington, and are otherwise sold by the publisher at its own price, which puts the actual text of federal design law beyond the casual reach of most of the people it governs.
The third is the gap between a measurement and a model output. Sample expansions, forecast scenarios and state-supplied estimates all produce numbers that look exactly like counts once they arrive in a press release. Separating them is not a matter of skepticism. It is a matter of reading documentation the publishing agency already wrote and almost nobody opens.
Reading the record a project leaves
The last thing in this section is the most concrete. Every highway project produces a set of drawings governed by a state plans manual, and those drawings encode decisions no summary preserves: which criteria the project was excused from meeting and who approved the exception, where the alignment was revised after stationing had been fixed, and which agency agreed to pay for which line item. The conventions carrying all of that are described in the guide to reading a plan set, which treats plan literacy as a research skill rather than a drafting one.
When a figure appears in this publication, its origin, its year and its method are stated, and where a figure cannot be verified the method for finding it is described instead. The reference material here exists so that the same standard can be applied to anyone else’s numbers.