The f2 similarity factor, in plain terms
Before a generic product can argue it behaves like the reference, it has to dissolve like the reference. The f2 similarity factor is the standard statistic regulators use to compare two dissolution curves - and 50 is the number that separates "similar" from "different".
What f2 actually measures
A dissolution test measures how much drug releases from a dosage form over time - typically the percentage dissolved at several time points, run in vessels of a defined medium. The result is a curve. A test product and a reference product each produce one, and the question is whether the curves are close enough.
The f2 statistic compresses that comparison into a single number from 0 to 100: a logarithmic transform of the average squared difference between the two curves at each time point. Identical curves score 100. Regulators - FDA and EMA alike - treat f2 of 50 or higher as "similar", meaning the profiles differ by no more than about 10% on average. Below 50 the profiles are considered different, and the argument for sameness weakens.
In practice the statistic is run per medium, not once: compendial media span pH 1.2 (fasted stomach), pH 4.5, and pH 6.8 (intestinal). A candidate that only matches in one medium can still behave differently where absorption actually happens - so a serious equivalence case demands f2 ≥ 50 across all compendial media, not just the friendly one.
| f2 value | Regulatory read | Practical meaning |
|---|---|---|
| 100 | Identical | Profiles coincide at every time point |
| 50 - 99 | Similar | Average difference within ~10%; supports sameness |
| Below 50 | Not similar | Release behavior differs; harder to argue equivalence |
Discriminating dissolution: making the test earn the number
A dissolution method that gives everything ~100% at 15 minutes cannot tell two formulas apart - it produces f2 near 100 regardless. A discriminating dissolution method is designed to expose the mechanism the formula depends on: the release-controlling layer, the dispersion's dissolution rate, the enhancer's contribution. If the test can distinguish a small change in composition, passing it means something.
This is why equivalence evidence in a dossier is not a single f2 value but a comparison package: reference profiles digitized across media, the candidate's simulated or experimental profiles overlaid on them, and the f2 result in each condition - with the method's discriminating power documented so a reviewer trusts the number.
The honest caveat: f2 is a release-behavior gate, not a guarantee of bioequivalence. Two products can match in vitro and still differ in vivo. It is the necessary first gate - the one that is cheapest to fail early in simulation, and most expensive to fail later in a clinic.