How to work out a food’s expiry date from the manufacture date

Some packages skip the calendar date entirely. Instead of "best before 03 2027," they print something like "manufactured 08 2026" alongside "best by 18 months from date of manufacture" — and leave the actual arithmetic to you. That is not a coded mystery like a Julian lot number. It is a real date, deliberately expressed as a starting point plus a duration, and it is meant to be calculated, not decoded.

Why some packaging does it this way

Printing a fixed calendar date means reprogramming the date-stamping machine for every single production run, on every product, in every market the item ships to. Printing a manufacture date plus a fixed duration instead means the duration can be set once, at the design stage, and never touched again — the stamping machine only ever has to fill in one changing number, the day it actually ran the line. For a product exported to a dozen countries with a dozen labelling conventions, or for large-format goods packed in bulk over a long production run, that is a meaningfully simpler system, and it is common on imported packaged food, spices, sauces, snack goods and anything sold in bulk or catering formats.

It also front-loads the honesty: a duration like "18 months from manufacture" is a single, stable promise about the product itself, made once and applied to every batch, rather than a fresh judgement call for each print run about what the calendar date should say.

Finding the manufacture date on the package

Look for a plain-language label first — "MFG," "manufactured," "production date," "packed on," or the equivalent in the package's other printed languages — usually followed by a date in day-month-year, month-day-year, or year-month-day order (which order depends on where the product was made, so check for a format hint nearby rather than assuming). That is the number this calculation starts from. It is a different thing from a lot or batch code: a lot code identifies a specific production run for recall purposes and does not reliably tell you a date at all, even when it is stamped right next to one. If what you are looking at is a short, unlabelled run of digits with no obvious calendar shape, it is very likely a batch code rather than the manufacture date, and decoding it is a different problem.

Doing the math, including the part that trips people up

Once you have a manufacture date and a duration, the calculation is addition — but it is calendar-month addition, not day-count addition, and the two do not always agree. "18 months from 08 2026" lands on 02 2028: count forward eighteen months, not 18 × 30 days. Where this gets genuinely tricky is a manufacture date that falls on a day near the end of a month, added across a duration that lands on a shorter month. A product manufactured on the 31st, with a six-month duration, is aiming for a month that may not have a 31st — and different manufacturers round that differently, some to the last day of the target month, some to the first day of the following one. When the exact day matters, that ambiguity is worth resolving conservatively: treat the earlier of the two plausible dates as the real one, rather than the later.

Where only a month and year are given for the manufacture date, do the same rounding the other pieces of this site cover for month-only dates: treat the manufacture date as the start of that month unless you have reason to think otherwise, which keeps your calculated result on the cautious side rather than the generous one.

What kind of date this actually is

A manufacture-date-plus-duration format is describing the same thing a printed best-before date describes elsewhere: a quality estimate for the sealed, unopened product, stored the way the label says to store it. It is not, on its own, a safety cutoff, and the reasoning that applies to any other best-before date applies here too — your own senses are reasonable evidence once you are close to or past the calculated date, for the categories where that is generally true. The one thing this format cannot tell you, any more than a plain printed date can, is how long the product is good for once it is opened. That clock is separate, starts the day you break the seal, and nobody printed it on the package in either format.

Making this not something you have to redo

• Do the calculation once, at the point of buying, not every time you wonder. A number written on the lid or the front of the package in marker is faster to read than the label's own printed text, and it does not require redoing the month arithmetic each time.
• Keep the manufacture date and the duration separate from any batch code on the same line. Only the plain-language-labelled number is the one this calculation uses.
• Treat the calculated date exactly like any other best-before date once you have it — a quality estimate with margin built in, not a hard line, and one that stops applying the moment the package is opened.

Munch does not ask anyone to do this arithmetic more than once. You log a batch when it comes into the house, with whatever date applies to it — calculated, printed, or estimated from memory — and the app tracks the countdown from there, including the separate, shorter one that starts the day you mark something opened, shared in real time with everyone in the household. The manufacture date on the package matters for exactly one calculation, at the start; after that, the date that matters is the one in the app.