Selling through distributors does not move the obligation, and not being able to see your state volumes is not the same as being under the threshold. This page covers the three allocation methods in order of defensibility, the two places in US law that expressly permit estimation, and what you have to write down for the estimate to hold up.
Free. Updated weekly. No login required.
Facts on this page were last verified against primary sources on September 9, 2026. The EPR Atlas is re-checked weekly; this stamp moves only when a verification pass actually runs, not when the site rebuilds.
Start here, because the question is usually asked in a way that assumes the answer. Selling through distributors, brokers or a wholesale channel does not shift producer status to them. In all seven enacted states the cascade stops at the first tier that exists, and a distributor tier sits at the bottom, reached only where no manufacturer, licensee, brand owner or importer can be identified. If you own the brand and your goods reach households in a state, you are the producer there whether or not you can see the sales.
California is the one place the channel can change the answer, and it is a definition rather than a loophole. PRC 42041(w)(5) provides that "for purposes of this chapter, the sale of covered materials shall be deemed to occur in the state if the covered materials are delivered to the purchaser in the state." It is drafted chapter-wide rather than confined to the producer definition, so it governs the small-producer threshold at PRC 42060 as well as the cascade. If you deliver to a distributor outside California and that distributor ships the goods in, the plain reading is that the sale was not delivered to your purchaser in California. The obligation would then land on the distributor under the seller and distributor backstop at PRC 42041(w)(3). Note the shape of that: it moves rather than disappears, no regulator has confirmed the reading, and it is the kind of position to document contemporaneously rather than assume.
Everywhere else, the practical fix is not a scope argument. It is a data clause in the distribution agreement, which is the subject of the last section on this page.
This is the trap that turns a data problem into a compliance problem, and it catches careful people.
Every tonnage test in every state is measured per state. So the sentence "we do not know our Minnesota volume" cannot be completed with "so we are probably under a ton there". You have to estimate the number before you can know whether the estimate clears the threshold, which means the work of estimating is not avoidable by being small. It is the thing that tells you whether you are small.
Two consequences follow. A company that files nothing because it could not measure has not claimed an exemption, it has simply not filed, and in California relief is not automatic in any case: it must be applied for and approved. And a company whose national volume is obviously large cannot assume the per-state split rescues it, because the thresholds are low. One ton is not much packaging.
Full thresholds and test logic are on the de minimis page.
These are not equally good and they are not interchangeable. Use the best one your data supports, and be able to say why the better one was unavailable.
One, ship-to state at order line level, from your own ERP. This is actual data rather than a proxy and it is what a regulator will expect a company with an ERP to be able to produce. If you have it, use it, and the rest of this page is a contingency plan rather than a method.
Two, retail sell-through or syndicated data. Retail point-of-sale where a customer provides it, or syndicated market data such as Circana or Nielsen. This is a measurement of where product actually sold rather than where you shipped it, which makes it better than ship-to for the redistribution problem below and worse for auditability, because you did not generate it and cannot always reproduce it.
Three, allocation of national volume. Population share is the usual basis, and it is a fallback rather than a method of choice. It is defensible only when documented as an allocation, applied consistently, and paired with a statement of what better data was unavailable and why.
Whatever you choose, changing basis between cycles without explaining why is worse than picking an imperfect basis and staying with it. An inconsistent series looks like an attempt to manage the number, and it is the pattern most likely to invite a question.
It is worth being precise here, because the distinction between what regulators accept in practice and what an instrument permits is exactly the distinction that collapses in most published guidance on this topic.
Two states permit it expressly, and they are the only two. Colorado, at 6 CCR 1007-2 Part 1 section 18.2.5(B), provides that a producer, if requested, must report the data necessary to meet its plan obligations "and may use prorated national data if state-specific data is not available or feasible to generate." Maryland, at COMAR 26.04.14.09D(2) and (3), uses the same words with one more condition: a producer "may use prorated national data if state-specific data is not available or feasible to generate upon approval from the Department."
Read the conditions, because they are on the face of both rules. Permissive rather than mandatory ("may use"), available only where state-specific data "is not available or feasible to generate", and in Maryland subject to MDE approval before it is used, not after. Neither is a default, and a producer that holds ship-to data in its ERP has a weak claim that state-specific data was not feasible to generate. Neither rule specifies the basis of proration. Population, revenue and unit share are all unnamed: none is blessed and none is forbidden.
The other five states are silent, and silence is not permission. No statute or rule in California, Oregon, Maine, Minnesota or Washington authorizes an estimation methodology. That does not make estimation improper, since a producer required to report a number it cannot measure directly has to derive it somehow, and Oregon's reporting guidance asks producers to document data sources, references, assumptions and calculations. It does mean that outside Colorado and Maryland you are relying on a documented, reasonable methodology rather than on a permission, and the page you write to defend it is doing the work that a citation would do.
If you take one thing from this section: cite Colorado 18.2.5(B) in Colorado and Maryland 26.04.14.09D in Maryland, get MDE's approval first in Maryland, and do not cite either anywhere else as though it travelled.
Separate from estimating total tonnage, and generally the bigger source of error in a filing.
When product ships to a distribution center in one state and sells in five, ship-to state is not state of sale. Using ship-to silently overstates the DC state and understates every state it serves, and the overstatement can be large enough to change which states you are above threshold in. Every enacted state is silent on how to handle this.
The defensible construction is a documented, reproducible redistribution basis: customer sell-through data where it is provided, retailer store counts in each state, or population share as the fallback, with the basis retained and applied the same way year over year. Note that this is a second methodology layered on the first, and it needs its own paragraph in the document described below rather than being folded silently into the tonnage estimate.
A foreign brand owner with no United States entity. Where no manufacturer, licensee or brand owner exists in the United States, the cascade continues. In Washington, Maryland, Minnesota and Maine it reaches the United States importer of record, and then the first distributor in or into the state. Colorado and Oregon reach the importer at their third tier on the same condition, that nobody in the tiers above exists within the United States. California has no importer tier at all and goes straight to whoever sells, offers for sale or distributes in or into the state. So an importing distributor of a foreign brand can be the producer, and often is.
Unbranded and generic product. Packaging that carries no brand identification is reached expressly at tier 1 in every state that has a manufacturer tier, which lands the obligation on the manufacturer rather than on anyone downstream. A distributor selling unbranded goods it did not make is generally not the producer; the factory is.
Both cases are worked through on Am I an Obligated Producer?, which also covers the four states where a written assignment can move the duty.
Under packaging EPR your distributor owes you nothing. It has no reporting duty of its own unless the cascade reaches it, and no obligation to hand you state-level sell-through data. Nothing in any of the seven laws creates one.
That is why this is a contracting problem rather than a regulatory one, and why it is fixable at renewal. The clause needs to specify the field (units or weight by ship-to or sell-through state), the cadence (annually, before your earliest state reporting deadline), the format, and a retention period long enough to survive an inquiry. Distribution agreements written before these laws existed rarely address any of the four.
Retention, since it decides how long the clause has to reach. The periods differ and none of the figures is widely known. Colorado requires producers to maintain records for five years (6 CCR 1007-2 Part 1 section 18.1.5(A)), with production due within ten business days. Maryland is also five years (COMAR 26.04.14.09B), with records available within fifteen business days or a longer timeline the Department sets (26.04.14.09C). California's is three years, and the duty as drafted runs to a PRO or independent producer rather than directly to every participant producer, so a producer inside a PRO is bound through its participant agreement rather than by the regulation itself (14 CCR 18980.4.2). Set the contractual period against the longest requirement you are actually subject to, not the shortest.
One document, refreshed annually, retained at least as long as your longest applicable retention period. It is the deliverable, and it is what separates a methodology from a guess.
It should state: what data sources were used and where they came from; why the chosen method was chosen, and what better method was unavailable and why; how any samples were selected and why they represent the portfolio; how supplier-provided weights were validated; the quality-assurance step; the state allocation basis and its rationale; every assumption and conversion factor with its source; and the date and the owner.
That list is assembled from Oregon's instruction to document data sources, references, assumptions and calculations, and the five-year retention periods in Colorado and Maryland. No single jurisdiction requires all of it, which is the point: a producer that has it will not lose an argument with any of them.
One rule that governs everything above. An estimate labelled as an estimate, with its basis stated, is a methodology. The same estimate presented as a measurement is a misstatement. Label it.
Work out whether you are the producer at all on Am I an Obligated Producer?, check whether your packaging is even in scope on B2B packaging and EPR, then run the numbers in the fee calculator and check your dates on reporting deadlines by state.
Every state guide carries the full fee schedule, registration deadlines, program plan status, eco-modulation detail, statute and rule text, and exemptions.