The federal government used to answer this question in regulation, and then it stopped. When FHWA published the rewritten National Bridge Inspection Standards at 87 FR 27396 in May 2022, the same rulemaking eliminated two older regulations: the Highway Bridge Replacement and Rehabilitation Program at 23 CFR part 650 subpart D, and the Discretionary Bridge Candidate Rating Factor at subpart G. The agency described both as outdated. Subpart D dated from 44 FR 15665 in March 1979, and it carried the federal definitions of rehabilitation, replacement and the sufficiency rating that bridge funding decisions had been written against.
The same year, the Specifications for the National Bridge Inventory discontinued the sufficiency rating as a reported data item, along with its asterisk field and the status codes for structurally deficient and functionally obsolete. Anyone still describing the choice between rehabilitation and replacement as a sufficiency question is describing a system that no longer files the number.
What the repealed rule said
It is worth reading, because the vocabulary survives in practice. Subpart D defined the sufficiency rating as the numerical rating of a bridge based on its structural adequacy and safety, essentiality for public use, and its serviceability and functional obsolescence. It defined rehabilitation as the major work required to restore the structural integrity of a bridge as well as work necessary to correct major safety defects. Under eligible projects it defined replacement as total replacement of a structurally deficient or functionally obsolete bridge with a new facility constructed in the same general traffic corridor, and required the replacement structure to meet the current geometric, construction and structural standards for the projected traffic over its design life.
That last clause is the reason replacement often wins arguments on paper. A rehabilitation restores what is there. A replacement has to be brought up to current standards, which means the comparison is rarely between two versions of the same bridge. It is between an old bridge made sound and a wider, straighter, differently loaded bridge, and only one of those two options changes the geometry of the road.
The GAO had already taken the sufficiency rating apart. In September 2008, in GAO-08-1043, the office recorded that FHWA assigns each bridge in the inventory a rating between 0 and 100 indicating its sufficiency to remain in service, where 100 represents an entirely sufficient bridge and 0 an entirely insufficient one, and concluded that the program built on it lacked goals focused on a clearly identified federal or national interest, performance measures, and sustainability. It also documented what states did with the money. From 1998 through 2006, 27 states transferred approximately $2.8 billion in program funds to other Federal-aid highway programs, which GAO put at about 7 percent of the $37 billion distributed over that period.
The vocabulary that replaced it
FHWA’s Bridge Preservation Guide now carries the operative definitions, and it sorts bridge work into four categories rather than two. Routine maintenance is work performed in reaction to an event, season or activity for short-term operational need with no preservation value. Preservation, also called preventive maintenance, prevents, delays or reduces deterioration and keeps bridges in good or fair condition, and splits into cyclical activities on predetermined intervals and condition-based activities responding to known defects. Rehabilitation is major work required to restore structural integrity along with work necessary to correct major safety defects, a definition the guide footnotes to the now repealed 23 CFR 650.403(c). Replacement is total replacement with a new facility in the same general traffic corridor.
One line in the guide’s federal funds eligibility table decides more than the definitions do. Routine maintenance is not eligible for federal funds, because 23 U.S.C. 116(d) requires states and owner agencies to properly maintain projects financed with federal aid. Cyclical maintenance, condition-based maintenance, rehabilitation and replacement all are eligible. The examples the guide gives of the ineligible category include snow removal and application of de-icing chemicals, debris and dead animal removal, and asphalt patching with no membrane on a concrete deck. The cheapest interventions that most directly cause and then conceal deck deterioration sit outside the federal-aid boundary, and everything on the expensive side of the boundary is shareable.
Condition sets the menu, not the choice
The guide maps its action categories onto inspection output twice. Against general condition ratings, codes 9, 8 and 7 point to preservation and cyclical maintenance, 6 and 5 to preservation and condition-based maintenance, and 4 or below to rehabilitation or replacement. Against AASHTO element condition states, state 1 points to cyclical maintenance, state 2 to cyclical or condition-based maintenance when cost effective, state 4 to rehabilitation or replacement, and state 3 to a genuine decision: condition-based maintenance, or rehabilitation when the quantity of poor exceeds a limit at which condition-based maintenance is not cost effective, or replacement when rehabilitation is not cost effective.
The guide is candid about how coarse the component ratings are for this purpose. While general condition ratings may help in providing general categorization of preservation, rehabilitation and replacement needs, it says, they are too broad for determining specific activity on a bridge, and if only those ratings are available then detailed scopes are necessary to determine the work activities and estimate cost. The rating decides which of the three programs a bridge belongs to. It does not decide what the crew does. How those ratings are produced in the first place is set out in how bridges are inspected and rated.
There is also a threshold effect nobody designed. The guide notes that when a bridge component enters poor condition, bridge preservation typically ends until that component is rehabilitated back into good or fair condition, or replaced. A component in condition 4 is not merely worse than one in condition 5. It has left the cheap program.
Strengthening is the option that gets left out
The choice is usually framed as two options, and there are three. FHWA’s April 2019 Report on Techniques for Bridge Strengthening, publication FHWA-HIF-18-041, exists because owners and engineers are looking for efficient and economical methods to repair their bridges and increase their live load capacity, methods that are cost effective and constructible and that yield solutions lengthening the service life of the structure. The report observes that over the two decades before it was written the vast majority of new innovations in bridge strengthening involved applications of fiber reinforced composite materials, and that its own predecessor syntheses of traditional methods date from 1997 and 1987.
Strengthening changes what the load rating says without changing the geometry, which makes it the option that resolves a posting problem while leaving a narrow bridge narrow. Whether that is a success depends entirely on whether the deficiency being solved was structural or geometric, and the repealed federal definitions bundled those two together under functional obsolescence, which is one reason they were repealed.
The failure mode is not choosing wrongly
It is choosing and then not executing. The National Transportation Safety Board’s interim report on the Pittsburgh frame bridge that fell in January 2022 records a rehabilitation that happened and a maintenance regime that did not. A rehabilitation project in 2009 replaced the structure’s downspouts and required all of its drains to be cleaned, and the inspection report for that year confirms the work was completed and describes the drains as clear. By 2011 the report notes some had clogged again, and the Board traced clogged drains and the associated maintenance need through every report from then to 2021. Over the 11 years before the collapse, it found, the maintenance actions the annual reports called for were not performed.
That sequence is the argument for reading rehabilitation and replacement as points on a single funding decision rather than as rival engineering philosophies. Both consume capital that comes with federal participation. Both are recorded, scoped and defended. Cleaning a drain is neither, and it is not eligible for federal funds. The economics of comparing the options across a service life, including the user costs of each work zone, belong to highway lifecycle cost analysis, and the aggregate consequence of a nation of these decisions shows up in the state of the bridge inventory and across more on bridge engineering.
What the 2022 repeal left behind is a decision with no federal arbiter. The categories are defined in guidance rather than regulation, the number that used to sort bridges into eligibility bands is no longer collected, and the boundary that still has legal force is the one deciding who pays for the cheapest work.