A plain-language walkthrough of the FDA/EMA f2 similarity factor: what it measures, the formula explained step by step, acceptance criteria, and how to use the f2 similarity calculator.
The standalone calculator has one table and nothing else to set up. Each row is one time point, with three boxes: the time in minutes, the mean % dissolved for the Reference product (Rt), and the mean % dissolved for the Test product (Tt). The Reference column comes first and the Test column second, and the (Rt โ Tt) column is always Reference minus Test.
The (Rt โ Tt) and (Rt โ Tt)ยฒ columns in the input table stay empty until you calculate, and a sum row appears under the table afterwards.
Above the table is a drop-down labelled Decimal Precision (required before calculating), with options from 0 digits to 5 digits. Nothing is selected when the page opens. If you press calculate without choosing, the calculator stops with a message and moves the cursor to the drop-down. Changing the precision after you have results re-runs the calculation straight away.
The setting changes only what you see. The sum, the square root and the logarithm are worked out with full precision whatever you choose, so f2 does not move when you change the display. See the question about 0 digits in the FAQs below, because at low precision the visible rows may not add up to the visible total.
| Situation | What happens |
|---|---|
| No decimal precision chosen | A message appears and the drop-down is focused. |
| The table has no rows | A message asks you to load a template or add a row. |
| Any box in any row is empty | The empty boxes are outlined in red, the row number is named in the message, and the cursor jumps to the first one. Fill it or remove the row. |
| A % value below 0 or above 120 | The box is outlined and a message names the time point to check. Values up to 120 are allowed because assays can read slightly above 100%. |
The calculator does not check that time points are unique. Two rows at the same time are both used in the sum and both counted in n, so keep each time to one row.
Here are mean results for a Reference and a Test tablet. Choose 2 digits of precision and enter the rows below:
| Time (min) | Reference (Rt) | Test (Tt) | Rt โ Tt | (Rt โ Tt)ยฒ |
|---|---|---|---|---|
| 10 | 31.50 | 29.10 | +2.40 | 5.76 |
| 20 | 52.80 | 49.60 | +3.20 | 10.24 |
| 30 | 68.40 | 66.00 | +2.40 | 5.76 |
| 45 | 83.70 | 80.90 | +2.80 | 7.84 |
| 60 | 92.60 | 90.20 | +2.40 | 5.76 |
The calculation table under the results then shows these rows, followed by the working rows:
| Working row | Value |
|---|---|
| Sum of (Rt โ Tt)ยฒ | 35.36 |
| Sum of (Rt โ Tt)ยฒ รท n, with n = 5 | 7.07 |
| 1 + Sum of (Rt โ Tt)ยฒ รท n | 8.07 |
| Square root of that | 2.84 |
| f2 = 50 ร log [100 รท 2.84] | 77.33 |
The medium-calc line under the result card reads f2 = 50 ร logโโ[100 / 2.84] = 50 ร 1.55 = 77.33. Because 77.33 is between 50 and 100 and n is 5, the banner reads Dissolution Profiles: SIMILAR, and the result card shows SIMILAR with the note "FDA/EMA criterion (f2 โฅ 50)".
There is no 15-minute row in this data, so the note under the table says no 15-minute time point was found and the 85% check could not be performed. f2 was calculated from all five rows.
To see the other outcome, replace the Test column with 22.0, 38.5, 52.0, 70.1 and 81.4. The squared differences now add up to 874.10 and f2 drops to 43.87, so the banner turns red and reads NOT SIMILAR.
After every calculation the calculator looks for a row at exactly 15 minutes and shows one of three notes:
None of these notes removes or changes any row. The calculator uses every time point you entered.
| Situation | Banner | Result card |
|---|---|---|
| 3 or more rows, f2 from 50 to 100 | Green, SIMILAR | SIMILAR |
| 3 or more rows, f2 below 50 | Red, NOT SIMILAR | NOT SIMILAR |
| Fewer than 3 rows | Blue info banner, insufficient time points | Inconclusive |
Both buttons need a completed calculation. If you press them before calculating, a message asks you to calculate first.
Copy Full Report copies a text report: a title line, the table (time, Reference, Test, difference, squared difference, separated by tabs so it pastes neatly into Excel), then the f2 value with its tag and the verdict text.
Print Report opens an editable A4 preview that fits on one page. It carries the calculation table with its working rows, the verdict banner with the f2 value and the formula, and the chart as an image. You type these details directly on the sheet before printing: company name, dissolution medium (buffer), Reference brand and batch, Test brand and batch, analysis date, instrument, and the Analyzed by and Checked by lines.
The logo box accepts a PNG, JPG or WebP image of up to 1 MB, between 30 ร 30 and 512 ร 512 px on each side. HEIC photos from phones are not accepted. Choose Save as PDF in the print dialog and switch off Headers and footers for the cleanest file. Esc or the Close button leaves the preview.
Enter the mean % dissolved for each product at each time point, for example the average of 6 or 12 units. The standalone calculator does not average anything for you. If you only have unit-level data and want the calculator to work out the means and the %RSD first, use the Comparative Dissolution Profile calculator instead.
The calculate button is always clickable, but if no precision is chosen the calculator stops and shows a message asking you to select 0 to 5 digits. The choice controls how every number is displayed: the table, the sum, the verdict text, the math log, the chart labels, the copied text and the printed report. The calculation itself keeps full precision.
Because only the display is rounded. With the example in this guide, the true differences are 2.4, 3.2, 2.4, 2.8 and 2.4, but at 0 digits the rows show 2, 3, 2, 3 and 2, while the sum of squares is still calculated from the real values (35.36, shown as 35). Choose 2 or more digits when you want to check the arithmetic by hand.
No. The differences change sign, but they are squared before they are added, so f2 is identical. Only the colour and sign shown in the (Rt โ Tt) column change. It is still worth entering the products in the right columns so the report is correct.
No. Identical profiles give exactly 100, and every difference between them pulls the value down. If you ever see a number above 100 somewhere else, check the data or the formula used.
It still calculates f2, but the verdict banner says there are not enough time points for regulatory use, the result card shows Inconclusive, and a note under the table asks you to add more points. The row counter on the data-entry table also reminds you when you have fewer than 3 rows.
The SR template loads six rows at 60, 120, 240, 360, 480 and 720 minutes, which is 1 to 12 hours. Time is always entered in minutes, so convert hours before typing if you build your own rows.
No. Every row you enter goes into the calculation. If your guidance allows only one mean at or above 85% for each product, leave out the extra points yourself.
The calculation, chart, copy text and print preview all run in your browser, and the calculator script does not send your entries to a server.
Educational use only. This guide explains how the standalone f2 calculator on this site works. It is not a validated system, and it does not replace your SOPs or the guidance your regulator expects you to follow. Please read the disclaimer for how this site should be used.