Describe a drug product, and this builds the release specification that follows from its form, its route, its dose and where it is going. Twenty one modalities, from a tablet through an antibody drug conjugate and an mRNA particle to an autologous CAR-T, against a library of 121 tests. Twelve launch regions, each with the pharmacopoeia that governs there. The ICH thresholds are computed from the dose you entered rather than quoted from a table, and every place the regions disagree is flagged, which is the part that actually costs time. It also checks that the product you described could exist: an autologous cell therapy in a multi-dose container has no second dose to withdraw, and a terminally sterilized one arrives dead. The modality settles what the thing can be presented as, so combinations that could not exist are grayed out rather than left selectable: pick an AAV gene therapy and the suppository, the patch and the tablet go with it.
This is a teaching instrument, not a regulatory document. It produces a starting point for a conversation with your quality unit, assembled from public guidelines. A real specification is negotiated with a regulator, justified by your own batch and stability data, and approved in your dossier. Nothing here has been reviewed by any agency, and nothing here should be filed, released against, or treated as advice.
Every row carries one of three tags, and knowing which is which is most of the skill in reading a specification.
compendial The criterion is fixed by the pharmacopoeia and is the same for everyone. Sterility is no growth. Uniformity of dosage units has an acceptance value of not more than 15.0 at stage one. Extractable volume is not less than the label. You do not negotiate these, you meet them.
ICH The criterion follows from an ICH guideline and a property of your product, usually the dose. The impurity thresholds, the mutagenic acceptable intake, the elemental and solvent concentration limits and the endotoxin limit are all in this class, and all of them are computed on this page from what you entered rather than copied from a table. Change the maximum daily dose and watch them move.
applicant The criterion is yours. No guideline says an assay must be 95.0 to 105.0 percent, or that aggregates must be below some figure. You propose a limit, justify it from batch history, stability and toxicology, and the regulator approves it or does not. Roughly half of a real specification is this class, and it is where the argument happens.
A tool that presented all three the same way would be teaching the wrong lesson, which is why the tags are on every row rather than in a footnote.
Most of the analytical chapters people expect to differ do not. Sterility, bacterial endotoxins, uniformity of dosage units, dissolution apparatus, microbial enumeration and elemental impurities are all harmonized through the Pharmacopoeial Discussion Group or adopted from the same ICH text, so USP, Ph. Eur. and JP say the same thing in different numbering. The tool shows all three references side by side anyway, because the numbering is what you have to write on the certificate.
The real divergences are fewer and sharper. Antimicrobial effectiveness is the clearest: Ph. Eur. 5.1.3 criterion A is stricter than anything in USP <51>, and a formulation that passes comfortably in the US can fail it, which is why EU requirements often set the preservative level for the world. Inhalation is the other: USP <601> and Ph. Eur. 2.9.18 are not harmonized, and a product filed in both regions usually carries both tests. Recombinant factor C arrived as a full alternative in Ph. Eur. 2.6.32 before the US had an equivalent chapter. USP <787> exists for therapeutic proteins and has no direct Ph. Eur. counterpart.
The largest differences are not analytical at all. A batch made outside the EU normally has to be tested again inside it before a Qualified Person can certify it, unless a mutual recognition agreement covers the country it came from. There is no comparable requirement in the US or Japan. That single rule shapes supply chains more than any acceptance criterion on the sheet, which is why the manufacturing location is one of the questions this page asks.
The regions divide into three kinds, and the table says which kind each one is rather than papering over the difference.
Regions that publish their own chapter numbering. The United States, the European Union, Japan and China. Those references are given as chapters, and for the first three the numbering is carried here.
Regions that recognize somebody else's pharmacopoeia. The United Kingdom's BP incorporates Ph. Eur. texts; Switzerland applies Ph. Eur. alongside a national pharmacopoeia; Canada recognizes several through Schedule B; Australia points at the recognized pharmacopoeias through Therapeutic Goods Orders. Those show as the local book followed by whose chapter it is, for example BP, per Ph. Eur. 2.9.40.
Regions whose numbering is not carried here. China, Brazil, India, Korea and the International Pharmacopoeia publish their own general chapters, and the numbering is not in this tool. Those rows say chapter to be confirmed rather than showing a number. That is deliberate. An invented chapter number in a document that looks like a specification is worse than an admitted gap, and the Chinese Pharmacopoeia in particular is not always equivalent to the USP or Ph. Eur. text even where the test carries the same name.
Modality specific tests mostly have no compendial chapter anywhere, because the technique is newer than the pharmacopoeias' coverage of it. Those show the method rather than a chapter, in every region, which is what a real specification does too.
What was checked, and how. Every chapter number carried here was checked against published sources rather than written from memory, and nine were wrong. The Japanese references were the worst: JP 2.60 is melting point and 2.61 is turbidity, not the electrophoretic methods they were attached to, and JP 4.03 is the digestion test, not preservative efficacy. Ph. Eur. 2.9.9 is penetrometry, not patch adhesion. Amino acid analysis, size exclusion and glycan profiling all have their own chapters in both books and were pointing at generic ones. Dynamic light scattering acquired a harmonized chapter in 2024 and the tool was citing an informational one. The residual solvents guideline moved to its ninth revision in January 2024. All of that is corrected. What could not be checked is the full text of any chapter, only its number and title, so a reference here tells you where to look and never what you will find when you get there.
Impurity thresholds, ICH Q3A and Q3B. Both scale with the maximum daily dose, and both take the lower of a percentage and an absolute daily intake. At a 500 mg daily dose the identification threshold for a degradation product is the lower of 0.2 percent and 2 mg, and 2 mg of a 500 mg dose is 0.4 percent, so 0.2 percent governs. Drop the dose to 5 mg and the 20 microgram limit becomes 0.4 percent, which now governs instead of the 0.5 percent. The page does that comparison for you and shows the winner.
Mutagenic impurities, ICH M7. The acceptable intake is a staircase in treatment duration: 120 micrograms a day for up to a month, 20 up to a year, 10 up to ten years, and 1.5 for longer. Dividing by the daily dose turns that into a concentration, which is what goes on the specification.
Residual solvents and elemental impurities, ICH Q3C and Q3D. A permitted daily exposure becomes a concentration when you divide by the daily intake of product. Option 1 in both guidelines assumes 10 grams a day, which is where the familiar numbers come from: methanol at 30 mg per day gives 3000 ppm, and cadmium at 5 micrograms per day gives 0.5 ppm. Set the daily mass on this page to 10 grams and you will get exactly those figures back, which is the check that the arithmetic is right.
Bacterial endotoxins. The limit is K over M, where K is 5.0 endotoxin units per kilogram for most parenteral routes and 0.2 for intrathecal, and M is the maximum human dose per kilogram in a single hour. The intrathecal case is twenty five times tighter, and it is tighter because the blood brain barrier is not in the way.
It does not know your product. It knows a dosage form, a route and a dose, and it applies the rules that follow from those. It cannot know that your molecule photodegrades, that your excipient reacts with the active, that your process leaves a specific residue, or that a reviewer asked for an extra test last time. Those are exactly the rows a real specification lives or dies on and they are exactly the rows this cannot generate.
The applicant-set limits shown as commonly this or that are conventions, not requirements, and they are there to show the shape of a criterion rather than to be adopted. Using one because it appeared here would be the wrong lesson entirely.
Chapter numbers move. Pharmacopoeias revise, ICH publishes new revisions, and a reference that was right when this was written can be superseded without anything on this page changing. Treat the references as a way to find the current text, not as the current text.
There is no monograph lookup. Where an official monograph exists for your substance or product, it governs, and its tests can differ from the general chapters used here. Nothing on this page has been checked against any specific monograph.
Nothing you type is sent anywhere. The product details, the results you enter and the certificate all stay in this browser, and the page has no server behind it to send them to.