Administrative records: building a case against EPA inaction

Under TSCA Section 21, a citizen may seek de novo judicial review of a petition denial. That clash over what a court may consider has shaped the path of Food & Water Watch v. EPA—and offers a practical lesson for anyone documenting agency inaction.

The phrase “administrative record” can sound like a filing cabinet with a legal label. In federal water litigation, it can become a contested perimeter around the case: what the agency received, what the court may examine, and who gets to put evidence before the judge. The stakes are concrete. A strong scientific argument can still be derailed by a weak evidentiary trail or a procedural fight over how the record was assembled.

The TSCA Section 21 framework: deadlines and agency inaction

Section 21 of the Toxic Substances Control Act gives citizens a route to ask EPA to begin rulemaking under Section 6. The statute gives the agency 90 days after receiving a petition to grant or deny that request. If EPA denies it, the statute provides for judicial review under 15 U.S.C. § 2620(b)(4)(A), including de novo review.

That deadline is a clock, not a guarantee of the result petitioners want. A denial within 90 days may still be challenged. An agency’s failure to act also needs to be documented carefully: the petition’s receipt date, the requested action, the agency’s response or lack of one, and the procedural steps that followed. Without those basics, a dispute about substance can become a dispute about what happened when.

For fluoride litigation, the factual issue at the center of the case concerns water fluoridation at 0.7 milligrams per liter, identified as the optimal concentration. The legal question under TSCA is whether the chemical presents an unreasonable risk of injury to health, and what action the statute requires if that standard is met. Those questions depend on scientific evidence, but the court’s review also turns on the record properly before it.

A petition is therefore both a request to an agency and the beginning of a potential court file. It should make the requested agency action clear, connect the evidence to the governing statutory standard, and preserve a traceable account of what was submitted. That is the foundation for documenting agency inaction on fluoride without asking a court to reconstruct the administrative history from scraps.

De novo review and the administrative record

The central distinction is easy to flatten and expensive to misunderstand. Judicial review of agency action often focuses on the materials the agency considered. TSCA Section 21, however, expressly provides for de novo judicial review of a petition denial. The provision does not mean that every new study automatically enters evidence, or that ordinary procedural rules vanish. It does mean that the review is not simply confined by a blanket rule to the initial agency record.

The litigation made that tension visible. In 2018, the Northern District of California denied EPA’s motion for a protective order that sought to limit review strictly to the initial administrative record. That ruling did not settle every later fight about evidence. It did, however, reject the idea that the initial record alone necessarily defines the limits of TSCA Section 21 review.

The distinction matters for anyone compiling an administrative record for TSCA lawsuits. Petitioners need to preserve the materials submitted to EPA, but they also need to think about evidence that may be relevant in court. Those are related tasks, not identical ones.

The table is a map, not a substitute for the statute or the court’s orders. De novo review does not mean “anything goes.” Evidence still has to be presented through proper channels, and the parties must have a fair opportunity to address it. Confusing the breadth of review with freedom from procedural rules is exactly how a case can lose time—and ground.

De novo review widens the court’s lens. It does not authorize anyone to move the evidence around without process.

Food & Water Watch v. EPA and the party-presentation principle

In September 2024, after bench trials, Senior U.S. District Judge Edward Chen ruled that fluoridation at 0.7 mg/L presents an unreasonable risk of injury to health under TSCA and ordered EPA to initiate regulatory risk management. That was a district court ruling on the merits. It was not the end of the case.

On May 21, 2026, the Ninth Circuit vacated and remanded the district court’s decision. Its ruling focused on procedure: the court held that the district judge had violated the party-presentation principle by taking control of the evidentiary record to include a study that had not been submitted by a party. The appellate court directed the district court to issue its ruling based on the evidence presented during the initial bench-trial record.

That holding must be read at its actual width. The Ninth Circuit did not declare water fluoridation definitively safe. It addressed who may present evidence and how the court may build the record. A procedural reversal can undo a judgment without resolving the underlying scientific dispute in the way either side would prefer.

For petitioners, the lesson is blunt: do not assume a judge will repair a gap in your evidentiary presentation. For agencies, the ruling is no license to treat Section 21 review as a narrow, closed-record exercise. The statute’s de novo provision remains part of the framework. The remand concerns the evidence used in this case and the court’s role in assembling it.

A robust legal evidence package for environmental health petitions has to survive both kinds of scrutiny. It must be capable of supporting the claim on the science, and it must arrive through a procedure that gives the opposing party notice and an opportunity to respond. A study that matters but was never properly presented can become the procedural vulnerability everyone argues about instead of the evidence itself.

Avoiding procedural traps in federal water litigation

The Ninth Circuit’s ruling puts party presentation in sharp relief. Parties identify the issues, submit evidence, and argue for the conclusions they believe the record supports. A judge decides the dispute on that presentation; the judge does not become an independent investigator responsible for filling holes in one side’s case.

That principle creates practical pressure points for litigation challenging water standards:

1. Keep the submission history exact. Preserve the petition and its supporting materials, along with reliable documentation of when EPA received them. Keep the agency’s denial and related correspondence with the same care. The statutory 90-day period makes dates central, not decorative.

2. Separate the scientific question from the procedural one. A claim about health risk under TSCA and a claim about whether EPA handled a petition lawfully may rely on overlapping material, but they are not interchangeable. State what each item of evidence is offered to establish.

3. Present evidence through the parties. Identify the studies and other materials on which the requested ruling depends. Follow the court’s directions for presenting them. A judge’s ability to understand the evidence does not cure a failure to submit it properly.

4. Track what changes after filing. New evidence may emerge while litigation is underway. Do not treat its existence as proof that it is already part of the record. The route for presenting it matters, especially when the opposing party has not had a chance to address it.

These points do not turn a complex federal case into a checklist exercise. They do expose where avoidable procedural disputes start. In litigation over EPA water standards, the record has a history: petition, response, pleadings, trial evidence, rulings, appeal. If that sequence is not kept straight, the argument can drift from the agency’s conduct to the court’s handling of proof.

Building the evidentiary foundation

The first job is to make the petition legible as a statutory request. It should identify the rulemaking action sought and explain how the supporting material bears on the TSCA standard. A collection of studies without a clear link to the requested action leaves EPA and, later, a court to guess at the theory.

The next job is provenance. For each material submitted, maintain a record of what it is, where it came from, when it was sent, and how it relates to the petition. For scientific studies, that means distinguishing the study itself from a summary or an advocate’s characterization of it. The distinction matters when a party later disputes what the evidence actually says.

Then preserve the agency’s response as part of the story. Section 21’s 90-day deadline gives the dispute a statutory timeline. The petition date, agency receipt, and denial date help show whether the agency acted within the prescribed period and what it decided. Do not quietly substitute a later court filing for a missing agency submission; the two serve different functions.

Finally, plan for contested evidence before trial. If a study may be central, make sure the parties have a procedural path to present it and address it. The 2026 remand makes the cost of improvisation plain. The district court’s 2024 merits ruling did not survive appellate review in its existing form because the court’s handling of one unsubmitted study crossed a procedural line.

None of this answers what EPA will ultimately propose after the remand proceedings. The specific risk-management rules remain unresolved. Nor does the record supplied by this case establish that future Section 21 petitions will follow a changed evidence-submission protocol. Those are open questions, and pretending otherwise would be the same kind of spin this litigation has spent years exposing.

The route through a TSCA fluoride case is narrow in a particular way: the law opens de novo review, while courtroom procedure still demands disciplined presentation. Build the petition so its logic can be followed. Preserve the paper trail. Put the evidence before the court through the parties, not by hoping the judge will find what counsel left out. Then watch the remand. In federal litigation, even a ruling on risk can be sent back to the file cabinet over who put one study on the table.