A restaurant holding waste in Poland keeps a waste record on an ongoing basis, and hands waste over with a transfer card that records what left, when, and to whom. The record lives in the national database on products, packaging and waste management. Assembling it once a year is not what the act describes, and the timestamps make that visible.
That last sentence is the whole page in one line. Venues that get into trouble here are rarely venues that refused to keep records. They are venues that kept them the way an invoice folder is kept: put in order the week before somebody asks. The Polish waste act is written against that habit, and it leaves marks, because the documents it describes carry clocks.
Article numbers on this page were read in the consolidated text of the waste act published by the Sejm — Dz.U. 2023 poz. 1587, ELI text type U, as at 23 July 2026. Where a value was not read, the page says so instead of guessing.
Restaurant waste is not just what leaves the dining room
A kitchen produces several distinct things that arrive in the same corridor. Mixed refuse from the dining room. The organic fraction from prep. Used frying oil, which has its own collector and its own paperwork. Glass, which in a bar can outweigh everything else by volume. Cardboard and film from deliveries. The occasional broken fridge or fluorescent tube from the storeroom.
Each has its own destination and, more importantly, its own counterparty. The moment two of them go to two different companies, you have two chains of paper, not one. Venues that describe their waste as "the bins" usually keep one chain and forget the second — most often the oil.
The act itself does not describe restaurants. It describes a holder of waste, and a restaurant becomes one the moment it produces waste. What the obligation covers, and what it does not, is set out in art. 66. The general duty to keep quantitative and qualitative records of waste, kept as you go and classified according to the waste catalogue, sits in art. 66 ust. 1. The exemptions sit in art. 66 ust. 4 — and one of them matters enormously for a small venue: producers of municipal waste are named there directly, in art. 66 ust. 4 pkt 1 lit. a. Another exemption, art. 66 ust. 4 pkt 4, points at types and quantities to be listed in a separate regulation issued under art. 66 ust. 5.
Waste codes are not printed on this page. The catalogue of waste is set by a separate regulation — art. 66 ust. 1 points to it explicitly, as rules issued on the basis of art. 4 ust. 3 — and that regulation was not opened for this article. Codes copied from memory are worse than no codes at all, because they look authoritative.
The record runs continuously, and that is what the act requires
The phrase in art. 66 ust. 1 that does the work translates roughly as "as you go". Not monthly. Not before an inspection. As the waste arises and moves.
Which is why the second half of the obligation is about documents rather than totals. Art. 67 ust. 1 pkt 1 lists what a holder of waste maintains: transfer cards under lit. a and waste record cards under lit. b, among others. The transfer card is about a movement between two parties; the record card is about a mass held by one party over time.
And art. 67 ust. 6 says where they are made: through the individual account in the national database. Not in a notebook, not in a spreadsheet on the manager's laptop. Documents produced through that account need no signature — art. 67 ust. 5 — which tells you what the legislator treats as the source of trust here. The account is the signature.
There is a deliberate exception for when the system itself is down. Art. 67 ust. 7 allows paper or electronic documents outside the database for the duration of an outage, art. 67 ust. 8 puts a notice of the outage on the administrator, and art. 67 ust. 9 requires the driver to carry the card during it. Then the data has to come back in: art. 67 ust. 10 and ust. 11 give no more than thirty days from the end of the outage. Thirty days is generous, and it is also a trap — the paper card ends up in a drawer, and a drawer is not a deadline anyone remembers.
There is also a lighter regime. Art. 71 allows a simplified record consisting of the transfer card alone, for producers below thresholds written into the article itself: hazardous waste up to 100 kilograms a year under pkt 1 lit. a, and non-hazardous, non-municipal waste up to 5 tonnes a year under pkt 1 lit. b. Whether your venue sits under those thresholds is a question about your venue, not about this page.
The transfer card: what it fixes, and why it carries a clock
The clock is in the article, not in the software
Art. 67 ust. 3a pkt 1 requires the transfer card to carry information about the date and hour of the start of the transport of the waste and of its delivery to the next holder. Date and hour. Both ends of the journey.
Consider what that means for a record kept "later". Nobody remembers a fortnight afterwards whether the truck came at 06:40 or 07:20 — and once you write a plausible hour you did not observe, you have not completed a record, you have manufactured one. The clock in the article makes the difference between recording and reconstructing visible in the data itself.
The card also carries the registration number of the entity. Art. 67 ust. 3 pkt 2 puts it there, referring to the number assigned under art. 54 ust. 1. And the record card, the other document, carries mass: art. 67 ust. 3b pkt 2 names the mass of waste produced, collected, received, handled in-house or handed over.
Retention is five years, counted from the end of the year
Art. 72 ust. 1 requires the documents and the data behind the record to be kept for five years, counting from the end of the calendar year in which they were produced — in practice up to five years and eleven months. Art. 72 ust. 2 requires them to be made available on demand to the bodies authorised to carry out checks.
The practical consequence is dull and important: whatever you use has to survive five and a half years, including the year you change accountants, change software, or lose the person who knew where things were. A folder on one laptop does not survive that, which is why connecting the tills, the bookkeeping and the stock into one place is a retention strategy rather than an IT preference.
BDO: the national database on products, packaging and waste management
The abbreviation stands for a name the act uses in plain text: Baza danych o produktach i opakowaniach oraz o gospodarce odpadami. Worth knowing the full form, because in conversation "BDO" is used loosely for three different things — the register, the account, and the reporting system — and they carry three different articles.
The register is kept by the marshal of the voivodeship: art. 49 ust. 1. Entry happens either on application, under art. 50 ust. 1, or ex officio for entities that already hold the relevant permits, under art. 51 ust. 1. Once a complete application arrives, art. 53 ust. 1 gives the marshal no more than thirty days to make the entry.
Two things follow from the entry. Art. 54 ust. 1 assigns an individual registration number — the same number that then has to appear in the record documents. Art. 55 ust. 1 creates the individual account in the database. That account is the thing your staff will actually log into.
And the register itself is not a separate creature: art. 234 ust. 1 states that the register referred to in art. 49 ust. 1 is created together with BDO. One system, one number, one account.
The annual report is a different object from the ongoing record
Art. 75 ust. 1 pkt 1 places the annual report on waste produced and waste management on the producer who is obliged to keep the record. Note the dependency — the report is built on top of the record, and it presupposes that the record exists. Art. 76 ust. 1 sets the deadline: by 15 March for the previous calendar year, filed with the marshal of the voivodeship, and under art. 76 ust. 1 pkt 2 by the place where the waste is produced.
| Feature | The ongoing record | The annual report |
|---|---|---|
| When it happens | continuously, as waste arises and moves | once a year, by 15 March for the previous year |
| What it consists of | transfer cards and record cards, art. 67 ust. 1 pkt 1 | a summary built from those documents |
| Where it is made | the individual account in the database, art. 67 ust. 6 | filed with the marshal of the voivodeship, art. 76 ust. 1 |
| Who is bound | the holder of waste, art. 66 ust. 1 | a producer obliged to keep the record, art. 75 ust. 1 pkt 1 |
| What it proves | that each movement happened and when | that the year's totals were declared |
| Failure looks like | gaps and implausible hours | a missing filing on a fixed date |
The act attaches penalties in several places. Failure to submit the application concerning the register carries, under art. 179, arrest or a fine. Failure to keep, make available or produce documents contrary to art. 72 carries a fine under art. 180 ust. 3. Failure to enter data into the database after an outage, contrary to art. 67 ust. 10 or ust. 11, carries a fine under art. 180 ust. 4. Failure to file the report required by art. 76 carries a fine under art. 180a.
No amounts are printed here, and that is the act's own doing. In the consolidated text those provisions carry the Polish formula "podlega karze grzywny" — and art. 179 "karze aresztu albo grzywny" — which is a fine, or arrest or a fine, with no sum named. Any figure at this point would be invented, and an invented figure in a legal article is the most expensive kind of mistake this page could make.
Who on shift makes the entry, and why it cannot be the accountant
The document has to be produced through the individual account (art. 67 ust. 6) and it has to carry an hour (art. 67 ust. 3a pkt 1). The person who knows the hour is the person who was standing by the back door when the truck arrived. That is a line cook, a kitchen porter, or a shift manager. It is not the accountant, who receives everything in a bundle at the end of the month and has no way of knowing whether the truck came at 06:40 or 07:20.
So the entry belongs to the shift, and it inherits every property of shift work: done under time pressure, by whoever is present, on a phone, between two other tasks. Design for that or it will not happen.
Three things make it survivable in practice:
- One named responsibility per shift, not per person. "Whoever is on the close" is a rule that survives staff turnover; "Marta does the waste" does not survive Marta's holiday.
- The entry happens at the door, not at the office. Every metre between the moment and the entry is a metre where the hour degrades into an estimate.
- The check is visible to the manager the next morning, not at month end. A missing entry found the next day can still be corrected honestly, because someone still remembers.
This is ordinary shift design, and it is the same problem as every other end-of-shift obligation. Making it a standing item that appears on the right shift rather than a thing people are told to remember is what a task layer is for, and giving it an owner inside the rota rather than a name on a list is what shift and people management is for. The wider question of which routines can be handed to a system at all, and which cannot, is worked through in what can actually be handed over to a system.
Used oil, glass, packaging: three streams with different fates
| Stream | Who receives it | How often | What proves it happened |
|---|---|---|---|
| Used frying oil | a specialised collector under its own contract | on a cycle tied to fryer volume | transfer card per collection, plus the collector's own confirmation |
| Glass | usually the municipal or contracted collector | on the municipal schedule | schedule and contract; a card where the stream is not municipal |
| Delivery packaging | contracted collector, or back to the supplier | daily inbound, batched outbound | transfer card, or the supplier's return note |
Fill that table with your own contracts and your own cards. Do not fill it with the numbers above — they are column headings, not data, and the whole point of the exercise is that the answers differ from venue to venue.
The stream nobody assigns is the one that disappears
The same asymmetry appears on the inbound side. The packaging that arrives with a delivery and the packaging that leaves with a takeaway order are governed by different rules and different obligations — the fee-and-deposit side of that is the subject of the takeaway packaging fee and deposit rules, and it is worth reading the two together, because the same cardboard box appears in both.
The gap between purchased weight and waste weight
The arithmetic is easy. Take the mass purchased in a period, subtract the mass sold and the mass formally written off, and look at what is left. The gap is real and it is measurable. What it is not is the mass of your waste.
A large part of the difference leaves the building as water vapour. Braising, roasting, reducing, boiling — a kitchen is, among other things, a machine for removing water from food. Trim and bone leave as waste; evaporated water does not leave as anything you can weigh. Any comparison that treats the gap as "waste" produces a number that is too big, in a stable and repeatable way.
Which is exactly why the number is still useful. A quantity with a constant systematic bias is useless as an absolute and excellent as a trend. Compare the gap to your own gap last month, in the same units over the same length of period, and the bias cancels. Compare it to somebody else's figure and you are comparing two different kitchens' evaporation rates.
The mass moving through the venue between purchase and plate is a separate measurement with its own logic, and it is worked through in inventory turnover in a restaurant. Between the two of them you can see whether a growing gap is a purchasing problem, a storage problem, or a portioning problem.
Where waste records and food loss meet, and where the two counts diverge
The waste record is a legal object. It counts what left the premises, by stream, with a counterparty and a moment attached, and it does not care why the material became waste. An over-produced tray of food and a tray of prep trimmings are, to the record, the same mass leaving through the same door.
The food loss count is a management object. It counts what stopped being sellable and, above all, why — over-production, spoilage, portioning error, breakage. Some of what it counts never becomes waste at all, because it gets redirected: staff meal, markdown, donation.
So the two numbers differ, and the difference is informative rather than embarrassing. If the loss count is far below the waste record, either losses are under-recorded or a lot of non-food mass is going out. If it is far above, material is being redirected before it becomes waste — which may be exactly what you want.
How to measure the second of those two properly, without turning it into a guilt exercise for the kitchen, is the subject of food waste in food service and how to measure it. Read that alongside this page and keep the two counts separate on purpose.
Three measurements that tell you whether the record is real
Everything above describes what the record should be. These three tell you what it is. All three are dimensionally honest, and all three can be run on paper before anyone builds anything.
Entry lag = system timestamp − hour on the card
Both sides are moments in time; a moment minus a moment is a duration, so the result is a length of time. The hour on the card is required by art. 67 ust. 3a pkt 1, which is what makes this measurable at all. Worked example with illustrative inputs: a card showing 06:40 and a system entry at 14:15 gives a lag of 7 hours 35 minutes, or 455 minutes. Run it across your last ten transfers and look at the shape of the distribution, not the average. Ten lags under an hour is a record. One lag of nine days is a reconstruction wearing a record's clothes.
Transfer completeness = cards issued ÷ actual collections
Both sides are counts of events, so the result is a dimensionless ratio. And here is the trap, stated out loud because it is the commonest way this measurement lies: if you count actual collections by counting cards, the ratio is 1.00 no matter how many collections went undocumented. The denominator has to come from somewhere independent — the contract schedule, the collector's invoices, the back-door log. Worked example with illustrative inputs: 9 cards against 11 collections established from the contract gives 0.82, and 0.82 × 11 returns 9 cards, which is the check that the arithmetic runs both ways.
Mass gap = mass purchased − mass sold and written off
Kilograms minus kilograms gives kilograms. All three quantities have to be in the same units over the same period, or the subtraction is meaningless. Worked example with illustrative inputs: 1 420 kg purchased against 1 180 kg sold and written off gives a gap of 240 kg. As established above, this is not your waste mass, and it should only ever be compared with your own figure from a previous period.
The reason to write all three down in this form is that each one names its own failure. A formula with its trap stated next to it is harder to misuse than a formula presented as a fact — which is broadly the argument in which numbers an owner actually looks at.
What to look at once a month so nobody reconstructs from memory
One: the lag distribution for the month. Not the average — the worst three. The average hides exactly the entries you need to see. If the worst three fall in the same week, ask what happened that week; it is usually a staffing gap, not a discipline problem.
Two: completeness against an independent denominator. Pull the collection schedule from the contract, not from the cards. Count what should have happened. Compare. Any month where the answer is exactly 1.00 deserves one moment of suspicion about where the denominator came from.
Three: the stream inventory. List the streams the venue produced this month and compare with last month. A new stream that appeared without a card — a refit, a broken freezer, a one-off clearance — is the most common single gap, and it is invisible to any measurement that only looks at streams you already knew about.
None of this needs software to start: the schedule, the cards, and fifteen minutes. Whether it later becomes an automatic summary rather than a manual one is a separate decision, and the honest version of it is set out under reports that assemble themselves and, for the venue as a whole, under end-to-end venue analytics.
What to have ready before the first inspection
The four things that take the longest to find
- The registration number, and where it lives other than in one person's inbox. It is assigned under art. 54 ust. 1 and has to appear in the record documents under art. 67 ust. 3 pkt 2.
- Account access held by more than one person. The account is created under art. 55 ust. 1; if only one login exists, the record has a single point of failure.
- The five-year archive, as an archive. Art. 72 ust. 1 counts from the end of the calendar year and art. 72 ust. 2 requires production on demand. The question is not "do we have it" but "can we hand it over this afternoon".
- The outage paperwork, if there ever was an outage. Art. 67 ust. 7 permits the paper card; art. 67 ust. 10 and ust. 11 require the data to be entered within thirty days of the outage ending. Paper cards that never made it back into the system are a specific, findable gap.
And one thing on no checklist: know which of your streams are covered and which are exempt. Art. 66 ust. 4 contains the exemptions, including municipal waste producers under pkt 1 lit. a and further types and quantities to be specified by regulation under pkt 4 and ust. 5. That regulation was not opened for this page, so this page does not tell you where your venue falls. Establish it with someone who has read it for your activity, and write the answer where the next manager will find it.
The same discipline runs through the whole legal cluster for a venue in Poland — approvals, hygiene, records — and its entry point is premises approval and HACCP. What the sector looks like in official statistics, so you can tell a venue-specific problem from a sector-wide one, is in Polish restaurant economics from official data.
Where to start this week
Take your last ten waste collections. For each one, put the hour written on the transfer card next to the timestamp of the entry in the system, and subtract. That difference — the entry lag — is the single measurement that separates a record from a reconstruction, and you can produce it this afternoon with the documents you already have.
Then get the denominator independently: how many collections should there have been, according to the contract rather than according to the cards. If those two numbers disagree, you have found something real before anyone asked you for it.
The rest of the venue's material is collected at the restaurant hub. Which system layer is worth adding once the routines are clear is worked through in CRM or ERP: which layer to add next, the four thresholds venues cross on the way there are in automation in a small business: four thresholds, and the restaurant-specific version — bookings, suppliers, reviews — is in restaurant automation.
Frequently asked questions
Does a restaurant have to keep a waste record in Poland?
The general obligation to keep an ongoing quantitative and qualitative record falls on the holder of waste under art. 66 ust. 1, and a restaurant becomes a holder the moment it produces waste. Exemptions sit in art. 66 ust. 4, including producers of municipal waste under pkt 1 lit. a and further types and quantities specified by regulation under pkt 4 and ust. 5. Where a particular venue falls depends on what it does, and that regulation was not opened for this article, so this page describes the mechanism rather than issuing a verdict.
How often is the waste record supposed to be updated?
Continuously. Art. 66 ust. 1 uses a phrase meaning "as you go", which is the opposite of a summary compiled later. The structural evidence is in the documents: the transfer card carries a date and an hour under art. 67 ust. 3a pkt 1, and an hour is not something anyone can honestly reconstruct weeks afterwards.
What does a waste transfer card actually record?
Under art. 67 ust. 3a pkt 1 it carries the date and hour of the start of transport and of delivery to the next holder, and under art. 67 ust. 3 pkt 2 the registration number assigned under art. 54 ust. 1. It documents a movement between two parties, unlike the record card under art. 67 ust. 1 pkt 1 lit. b, which under art. 67 ust. 3b pkt 2 carries mass produced, collected, received, handled in-house or handed over.
What is BDO and where is it named in the act?
The full name used in the act is Baza danych o produktach i opakowaniach oraz o gospodarce odpadami. Record documents are made through an individual account in it under art. 67 ust. 6; the account is created on entry into the register under art. 55 ust. 1; and art. 234 ust. 1 states that the register referred to in art. 49 ust. 1 is created together with BDO. The name appears in several articles rather than in one defining place, so the honest way to cite it is to point at the article carrying the norm you need. All of those provisions sit in the consolidated text of the act.
Who in the venue should be entering the records?
Whoever was present at the moment. The entry has to carry an hour, and only the person at the door knows it. In practice that means the shift, with one named responsibility per shift rather than per person, and a check visible to the manager the next morning rather than at month end. It cannot sensibly be the accountant, who receives everything in a bundle and has no independent knowledge of when the truck arrived.
Is the annual report the same thing as the ongoing record?
No, and treating them as the same is the most common expensive confusion here. The report under art. 75 ust. 1 pkt 1 is filed by 15 March for the previous calendar year with the marshal of the voivodeship under art. 76 ust. 1. It is built on top of the record and presupposes it. Filing a report does not retroactively create a record, and the act attaches separate consequences to the two failures — art. 180a for the missing report, art. 180 ust. 3 for documents not kept or not produced under art. 72.
Why does the weight of purchases never match the weight of waste?
Because a substantial part of the mass leaves as water vapour during cooking, and vapour cannot be weighed at the back door. The gap between purchases and what was sold and written off is therefore always larger than the waste that physically leaves. The bias is systematic and fairly stable, which makes the figure useless as an absolute and genuinely useful as a month-on-month trend against your own history.