Agricultural water is the part of the Produce Safety Rule that has taken the longest to settle and generated the most confusion. This page sets out the structure as it stands, what each section asks of a grower, and how a water treatment decision is actually documented for an inspector or an auditor. It is a guide to reading the rules, not a substitute for reading them — the text changes, and your operation’s obligations depend on facts we do not know.
About claims on this page
EcoloxTech generators produce a hypochlorous acid solution on site. They are regulated by the EPA as devices, are not registered disinfectants, and we make no efficacy claims for them or for the solution they produce. The only EcoloxTech product with disinfection claims is HOCl-AS200 (EPA Reg. No. 101112-1), used according to its label. Research cited here describes published studies of hypochlorous acid and electrolysed water under their own test conditions; it is not a claim for any product.
The shape of the requirement
Three sections do most of the work, and they operate in sequence: assess, meet the criterion where it applies, treat if you need to.
| Section | What it asks | What it produces |
|---|---|---|
| 21 CFR 112.43 | A written agricultural water assessment, reviewed at least annually and whenever conditions change, with mitigation decisions where the assessment identifies a hazard | A dated written document and a record of decisions |
| 21 CFR 112.44 | The microbial quality criterion for specified uses, and testing where required | Test results and a comparison against the criterion |
| 21 CFR 112.46 | If you treat the water, that the method is effective, delivered consistently, and monitored adequately | Set points, delivery records and monitoring records |
The 112.43 assessment is where most of the thinking happens, and it is broader than water chemistry. It considers the water system itself, the source, the degree of protection, practices and adjacent land uses, the crop, the method and timing of application, and the environmental conditions. It is written, it is dated, it is reviewed at least annually, and where it identifies a hazard it records what you decided to do about it. An inspector asking about your water will ask for this document first.
The criterion at 112.44(a)
For a defined set of uses, the rule sets a hard microbial quality criterion: no detectable generic Escherichia coli in 100 mL of agricultural water. The uses it applies to are the ones with direct contact at the highest-risk moments:
- Water used to wash or cool covered produce
- Water applied to harvested crops before cooling
- Water used to make ice that contacts covered produce
- Water contacting food-contact surfaces
- Water for hand washing during and after harvest
- Sprout irrigation water
Two consequences follow. First, untreated surface water may not be used for those purposes — the criterion is not achievable from an untreated surface source and the rule does not pretend otherwise. Second, meeting the criterion is a testing question as well as a treatment question: you demonstrate it with results, at a frequency your programme defines and can defend.
Pre-harvest irrigation sits under different provisions depending on the method, the crop and the interval before harvest. That is genuinely more complicated than the post-harvest picture, and it is the area where the rule has seen the most revision — which is exactly why the current text, not a summary of it, is what you should work from.
112.46, and why it matters to this discussion
The Produce Safety Rule names the device pathway
21 CFR 112.46(a) says that any method used to treat agricultural water — “such as with physical treatment, including using a pesticide device as defined by the U.S. Environmental Protection Agency (EPA); EPA-registered antimicrobial pesticide product; or other suitable method” — must be effective to make the water safe and of adequate sanitary quality for its intended use, and meet the microbial quality criterion in 112.44(a) where that applies.
In other words, a pesticide device is an expressly contemplated way to treat agricultural water under the rule. An onsite hypochlorous acid generator is such a device. What the rule then requires of you is operational rather than promotional: deliver the treatment consistently (112.46(b)) and monitor it with an adequate method and frequency (112.46(c)).
For a grower the practical importance of 112.46 is that it converts a marketing question into an operational one. It does not ask you to produce a label claim for your treatment. It asks three things:
| Requirement | What it means in practice | Evidence an inspector expects |
|---|---|---|
| 112.46(a) — effective method | The method you chose is appropriate to make the water safe and of adequate sanitary quality for the intended use, and meets 112.44(a) where that applies | Your rationale for the method, plus the test results showing the criterion is met |
| 112.46(b) — consistent delivery | The treatment is actually delivered throughout the water in a way that keeps it consistently safe | Set points, dosing records, alarm and deviation records |
| 112.46(c) — adequate monitoring | Monitored using an adequate method at an adequate frequency | The monitoring procedure and its completed records |
Where a documented, label-backed claim is required — by a buyer specification, a customer scheme, or another rule — that is a separate requirement and a registered product is the answer to it. HOCl-AS200 (EPA Reg. No. 101112-1) is used according to its label for that purpose. The two pathways coexist; the mistake is assuming one substitutes for the other in both directions.
The other FDA provisions
Three provisions outside the Produce Safety Rule govern what you may put in the water and at what level.
| Provision | Scope | Limit and conditions |
|---|---|---|
| FCN 1811 | Hypochlorous acid in process water or ice contacting fruits and vegetables | Free available chlorine not exceeding 60 ppm; drain or potable rinse for product consumed raw |
| 21 CFR 173.315 | Chlorine compounds in washing or lye-peeling fruits and vegetables | Potable rinse required |
| 21 CFR 178.1010 | Sanitising solutions on food-contact surfaces | Not more than 200 ppm available chlorine, with adequate draining before contact with food |
The most common internal error we see is a single site-wide ppm number applied to both water and surfaces. It is wrong in one direction or the other every time, and an auditor who knows the provisions will find it in the procedure before they find it in the plant.
Worker protection and FIFRA
Two regulatory frames apply to the people rather than the produce. Under FIFRA, an onsite generator is a pesticide device and the solution it produces is not a registered pesticide — which means no efficacy claims may be made for it, by us or by you. That constraint travels with the equipment: a grower who writes “kills 99.9% of pathogens” into a food safety plan for a generated solution has created a problem that did not previously exist.
Where a registered product is used, its label governs, and the label is the law: use sites, rates, PPE, re-entry and storage. Worker Protection Standard obligations attach to agricultural pesticide use where applicable, and a registered antimicrobial used in a covered way brings the training, notification and PPE elements of that programme with it. Also in scope is OSHA hazard communication for whatever concentrated chemistry remains on site — which, for an operation that has moved to onsite generation, is usually a considerably shorter list, and that is itself a defensible reason for the change.
GAP and buyer audit schemes
Most growers experience their regulatory obligations through a buyer audit rather than an FDA inspection. USDA Harmonized GAP, GlobalG.A.P., PrimusGFS and customer-specific schemes sit on top of the federal rules and generally ask for more documentation, not different practice.
| What auditors ask about water | What satisfies them |
|---|---|
| Where does your water come from, and has that been assessed? | The written 112.43 assessment, dated, with the annual review evident |
| What are your test results? | A testing schedule, the laboratory results, and a comparison against the criterion — with the out-of-specification results included, not filtered out |
| If you treat it, how, and how do you know it worked? | The method, the set points, the monitoring procedure, and completed monitoring records |
| What happens when a result is out of specification? | A written corrective action procedure and evidence it has actually been used |
| Who is trained on this? | Training records naming individuals and dates, including seasonal staff |
| Show me today’s records. | The current log, filled in, in the location the procedure says it lives |
One piece of advice repeated by every auditor we have spoken to: keep the deviations. A monitoring log with no out-of-specification readings across a season is not evidence of a well-run system, it is evidence that nobody is really measuring. A log that shows a deviation, a corrective action and a return to specification is the strongest thing you can show.
Organic production
If the operation is certified organic, or intends to be, the question of chlorine materials belongs with your certifier before any equipment decision. Chlorine materials are permitted in organic production only under specific conditions and limits, and the interpretation of those conditions — which water, which point in the process, what residual at what point — is the certifier’s to give. We cannot answer it for your operation, and neither can any equipment supplier. Ask early; the answer changes the specification.
Building the file
What a well-organised water file contains, in the order an inspector will want it:
- The written 112.43 agricultural water assessment, dated, with the annual review recorded
- A water system description: sources, distribution, treatment points, and where each stream is used
- The testing plan, and the laboratory results including any that failed
- The treatment procedure: method, set points by application, and the basis for them
- Monitoring records: what was measured, when, by whom, and what the reading was
- Corrective action records, with the deviations kept rather than removed
- Equipment records: calibration of test instruments, service on the generator, cell replacement
- Training records by name and date, including seasonal workers
- Safety data sheets for whatever chemistry remains on site
- The label for any registered product used, and the records its label requires
The implementation page turns this list into the procedures and logs that generate it.
The six findings we see most often
Across farm water programmes, the same small set of problems recurs. None of them is exotic and all of them are cheaper to fix before an audit than during one.
- One ppm number for the whole site. Process water and ice sit under FCN 1811; food-contact surfaces sit under 21 CFR 178.1010. One number cannot be right for both.
- An assessment that was written once. 112.43 expects at least an annual review and a review when conditions change. An undated assessment from three seasons ago is a finding on its own.
- Measuring in the wrong place. Irrigation residual read at the skid; wash residual read at the inlet. Both describe a system nobody is running.
- Logs with no deviations. A clean season of readings reads as unmeasured, not as controlled.
- No defined product disposition. Everyone knows to stop the line; far fewer have written down what happens to the product that already went through.
- Training records that omit seasonal workers. The people doing the work during the weeks that matter are frequently missing from the file entirely.
A morning spent against this list, once a year, is the cheapest audit preparation available.