The federal government has spent billions cleaning contaminated sites, but how much of that money can the Environmental Protection Agency recover from parties legally responsible for the pollution? An America Thinking comparison of EPA's published cost-recovery policies and a new Government Accountability Office investigation identifies an important gap: the agency explains how individual cleanup charges should be documented and pursued, while federal auditors say it lacks the broader measures needed to evaluate how effective recovery efforts have been across the program.

EPA's public cost-recovery guidance describes a straightforward principle. When the Superfund trust finances eligible cleanup work, EPA tries to recover appropriate costs from potentially responsible parties. The agency's documentation requirements include direct spending such as site work, employee time and contractors, along with indirect expenses and allocated overhead. Those rules matter because proving the amount and basis of a claim is different from establishing that a particular company has a legal obligation or sufficient resources to pay it.

The watchdog's new report examines outcomes over substantially longer periods. GAO says EPA spent approximately $25.8 billion at nonfederal National Priorities List sites during fiscal years 1996 through 2024, the period for which the investigators considered the expenditure data reliable. Across those sites, GAO estimates that recovered cleanup costs amounted to less than half of spending. That result is not a statement that the remaining dollars were all recoverable, lost through wrongdoing, or due immediately.

The difference is critical. Some contaminated sites have no identifiable responsible party, while other parties cannot pay or face liability disputes. The Superfund program exists partly because urgent public-health remediation may need to happen before a responsible party is found or a legal claim is resolved. EPA's policy itself says the strength of liability evidence, financial capacity and expected recoverable amount matter when deciding which recovery actions to pursue.

The GAO also distinguishes the financing of Superfund from the recovery of specific costs. It reports approximately $34 billion in constant 2024 dollars of trust-fund receipts during fiscal years 2006 through 2024, much of it from general federal funding, and nearly $32.8 billion appropriated from the trust during that period. Those figures do not measure site-level recovery success and should not be subtracted from one another to calculate unrecovered cleanup charges.

What remains uncertain is whether the government's enforcement effort is optimally directed. The GAO found EPA has performance measures for taking timely actions, but not comparable tools to evaluate the effectiveness of recovery across nonfederal priority sites. Agency officials explained that pursuing recovery is a continuing process that can take years. That is a legitimate timing complication, but it also makes a carefully defined long-run scorecard more important if policymakers want to compare regional practices or allocate enforcement resources.

This comparison highlights a distinction between procedural compliance and public accountability. EPA can document an invoice, conduct a liability search and open a recovery action without that evidence showing what share of potentially collectible spending will ultimately be reimbursed. Conversely, a low observed recovery rate does not establish poor performance unless the universe of legally collectible costs and the time needed to collect them are understood. Both the records and the limitations must be visible.

The GAO describes staffing constraints that complicate the task. It cites a reduction in the number of experienced staff who identify potentially responsible parties over the preceding 10 to 15 years, alongside legal and practical limits on collecting costs. EPA's enforcement materials emphasize pursuing appropriate companies and negotiating cleanup or reimbursement agreements. Reading them side by side suggests a concrete research question: which investigative approaches actually recover more money while preserving fair allocation of liability?

An effective public accounting would separate the total cost of remediation, costs legally attributable to identified parties, recoveries already received, claims still pending, amounts subject to settlements or bankruptcy, and expenses for which no liable or solvent party can be identified. It would also disclose cohort dates so older cleanup projects are not unfairly compared with recoveries from recent actions that remain open. The government records reviewed here do not provide that integrated performance table.

This is an original cross-document analysis of EPA's own enforcement standards and a GAO examination of measured recovery results. America Thinking did not investigate an individual Superfund site, inspect EPA enforcement files, or interview potentially responsible companies. The supported finding is limited but consequential: existing rules document how recovery is pursued, yet the oversight record identifies an agency-wide measurement gap affecting billions in environmental cleanup spending.

The practical follow-up is to ask EPA what consistent data it can provide about eligible recovery claims and results by fiscal year and region, what share remains under active collection, and how it will judge its own efforts over time. Better measurements could inform both taxpayer oversight and remediation policy without compromising the public-health purpose of the cleanup program.