CFU Calculator (Colony-Forming Units per mL)

Background.

A CFU calculator converts colonies counted on an agar plate into colony-forming units per millilitre of the original sample. The arithmetic is short — colonies divided by the volume of original sample that reached the plate — but the two things that decide whether the answer is defensible are the plate you chose to count and the number of digits you report, and this calculator handles both explicitly.

For a single plate the relation is CFU/mL = C ÷ (V × d), where C is the colonies counted, V is the volume plated in millilitres, and d is the dilution expressed as a fraction (10⁻⁵, entered here as the exponent −5). The product V × d is the volume of original, undiluted sample that actually reached that plate. Plate 0.1 mL of a 10⁻⁵ dilution and you have delivered 10⁻⁶ mL of original sample, so 68 colonies means 68 ÷ 10⁻⁶ = 6.8 × 10⁷ CFU/mL.

When two consecutive ten-fold dilutions both give countable plates, discarding one of them throws away real information. Both ISO 7218 and the FDA's Bacteriological Analytical Manual therefore prescribe a weighted mean: N = ΣC ÷ [V × (n₁ + 0.1 n₂) × d], where ΣC is every colony on every retained plate, n₁ and n₂ are the plate counts at the first and second dilutions, and d is the first (lower) dilution. The 0.1 weight is not a fudge factor — a plate at the next decimal dilution carries one tenth as much original sample, so it earns one tenth the weight, and the denominator is again simply the total volume of original sample across all the plates.

That formula was checked here against two independent authorities that do not cite each other, and they agree exactly. FDA BAM Chapter 3 §D.1.a prints the example 232 and 244 colonies at 1:100 plus 33 and 28 at 1:1000, giving 537 ÷ 0.022 = 24,409, reported as 24,000. The ISO/TC 34/SC 9 committee's own public verification report of the ISO 7218 calculator prints 1.7 × 10⁴ for 168 and 14 colonies at 10⁻² and 10⁻³, and 1.9 × 10⁴ for 168 + 215 and 14 + 25. This calculator reproduces all four numbers, and each is a permanent test.

Where the authorities do not agree is the countable range, so it is an editable input rather than a hard-coded constant. FDA BAM says 25–250 colonies per plate and reports anything outside that as an estimated count. APHA Standard Methods uses 30–300 for heterotrophic plate counts in water. The ISO method standards set it per method rather than with one universal number. The defaults here are FDA BAM's, because that is the range that could be read verbatim from a primary document, and the alternatives are named so you can change two fields and work to your own standard.

Reporting is the other half of the job. FDA BAM §D is unusually explicit: 'To avoid creating a fictitious impression of precision and accuracy when computing APC, report only the first two significant digits.' It even specifies the tie-breaking rule — round up when the third digit is 6 to 9, down when it is 1 to 4, and when the third digit is 5, up if the second digit is odd and down if it is even, which is round-half-to-even. This calculator returns the full-precision value and the correctly rounded reported value side by side, so you round once and can see exactly what rounding cost you. BAM's four published examples (12,700 → 13,000; 12,400 → 12,000; 15,500 → 16,000; 14,500 → 14,000) are boundary tests here.

One behaviour is deliberate: a plate with no colonies is refused rather than answered with zero. Nothing growing does not mean nothing is there; it means the concentration is below the method's detection limit, which is 1 ÷ (V × d). The calculator supplies that value so you can report the correct '< X CFU/mL' statement, exactly as ISO's own worked example prints '<1.0E+01' and as FDA BAM §C.4 instructs. Finally, remember what the unit means: a colony-forming unit is one thing that grew into one colony on the medium you offered, at the temperature and time you offered it. A chain of ten streptococci is one CFU. A cell that is alive but not culturable on that plate is invisible to the method entirely.

What is cfu calculator?

A colony-forming unit is the operational unit of a viable plate count: one discrete colony on an agar plate is taken to have arisen from one CFU in the plated aliquot. The deliberately awkward name exists because a CFU is not the same as a cell. A clump, a chain or a pair of cells that stays together through plating grows into a single colony and counts as one CFU, so a viable count systematically under-reads cell number for chain-forming and clumping organisms. Conversely, a cell that is alive but cannot grow on the medium, temperature or atmosphere you offered contributes nothing at all — the viable-but-non-culturable state is a well-documented blind spot of the whole approach.

The measurement is therefore defined by its conditions, which must travel with the number. FDA BAM Chapter 3, for example, specifies plate count agar tempered to 45 ± 1 °C, 12–15 mL poured within 15 minutes of the original dilution, and incubation at 35 ± 1 °C for 48 ± 2 h — 32 ± 1 °C for milk. A CFU/mL quoted without its medium, temperature and incubation time is not reproducible, because 'colony-forming' is only defined relative to what was on offer.

The countable range exists for two opposing reasons. Below the lower bound the sampling statistics are poor and the dilution factor multiplies that noise: 5 colonies from a 10⁻⁶ dilution has a Poisson error near 45%, and the factor of a million amplifies that error into the final answer without changing its relative size. Above the upper bound, colonies merge, are hard to resolve and mutually inhibit growth by competing for nutrients and by producing metabolites, so a crowded plate systematically under-reads. FDA BAM puts it plainly: counts outside 25–250 'may give erroneous indications of the actual bacterial composition of the sample'.

Because of both effects, results outside the range are reported as estimated counts rather than as measurements, and results derived from a plate with nothing on it are reported as a '<' value rather than as zero. Those conventions are not pedantry; they are the difference between a number that survives an audit and one that does not.

How to use this calculator.

1. 01Prepare decimal dilutions and plate them. FDA BAM makes 10⁻², 10⁻³, 10⁻⁴ by transferring 10 mL of the previous dilution into 90 mL of diluent, then pipettes 1 mL of each into duplicate plates — use the serial dilution calculator to plan the chain.
2. 02Incubate to the schedule for your method, then count every colony including pinpoint ones. Record which dilution each plate came from.
3. 03Choose the mode. If only one dilution gave a plate inside the countable range, use single plate. If two consecutive ten-fold dilutions both did, use the weighted mean — it uses both, which is what ISO 7218 and FDA BAM require.
4. 04Enter the volume you plated per plate (1 mL for a pour plate, 0.1 mL for a spread plate) and the dilution as a power of ten: −5 for 10⁻⁵. In weighted-mean mode, enter the FIRST (lower) dilution; the second is assumed to be the next ten-fold step.
5. 05Set the countable range to match your standard: 25–250 for FDA BAM, 30–300 for APHA water methods, or whatever your specific ISO method prescribes.
6. 06Read the countability note before the number. If a dilution falls outside the range, the result is an estimated count and should be labelled as one.
7. 07Write down the reported value, not the full-precision one — two significant figures, per FDA BAM §D. Keep the full-precision value only if you are feeding it into a further calculation such as a log reduction.
8. 08If nothing grew on any plate, do not report zero. Report the detection limit as a '<' value — the calculator gives it as 1 ÷ the effective volume.

The formula.

Both forms are the same statement: colonies divided by the volume of original sample they came from.

For one plate, you pipetted V millilitres of a d-fold dilution, so the original-sample volume that landed on the agar is V × d millilitres. Plating 0.1 mL of a 10⁻⁵ dilution delivers 0.1 × 10⁻⁵ = 10⁻⁶ mL of original sample, and 68 colonies from that is 68 ÷ 10⁻⁶ = 68,000,000 CFU/mL, or 6.8 × 10⁷ — which is already two significant figures, so the reported value is unchanged.

For two consecutive dilutions, the same logic applies to every plate at once. Each plate at the first dilution carries V × d millilitres of original sample; each plate at the next ten-fold dilution carries one tenth of that, V × 0.1d. Adding them up over n₁ plus n₂ plates gives a total of V × (n₁ + 0.1 n₂) × d millilitres, and dividing the pooled colony count by that total is the weighted mean. Written out, N = ΣC ÷ [V × (n₁ + 0.1 n₂) × d]. FDA BAM omits V from its version of the formula only because its pour-plate method always plates exactly 1 mL; setting V = 1 here reproduces BAM's expression exactly, which the test suite asserts.

Working through FDA BAM's own published example: 232 and 244 colonies at 1:100, 33 and 28 at 1:1000, 1 mL per plate. ΣC = 232 + 244 + 33 + 28 = 537. The denominator is 1 × (2 + 0.1 × 2) × 0.01 = 2.2 × 0.01 = 0.022 mL of original sample. So N = 537 ÷ 0.022 = 24,409.09…, and BAM prints '= 537/0.022 = 24,409 ≈ 24,000'. The ISO 7218 examples work identically: 168 and 14 colonies at 10⁻² and 10⁻³ with one plate each gives 182 ÷ (1 × 1.1 × 0.01) = 16,545.45 → 1.7 × 10⁴, and 168 + 215 with 14 + 25 gives 422 ÷ 0.022 = 19,181.82 → 1.9 × 10⁴.

The detection limit falls straight out of the same denominator: one colony would represent 1 ÷ (V × (n₁ + 0.1 n₂) × d) CFU/mL, so that value is the smallest concentration the plating scheme can distinguish from nothing, and it is the '<' figure to report when no colonies grow.

Rounding stage — and there are two of them here, deliberately. The computed titre is carried at full working precision and rounded only at the return boundary to twelve significant figures; nothing intermediate is rounded, so the base-10 logarithm is taken from the unrounded value. The reported titre is a separate output rounded to exactly two significant figures, because FDA BAM §D requires it and specifies how ties break: 'Round by raising the second digit to the next highest number when the third digit is 6, 7, 8, or 9 … Round down when the third digit is 1, 2, 3, or 4. When the third digit is 5, round up when the second digit is odd and round down when the second digit is even.' That last clause is round-half-to-even, and BAM's four worked examples — 12,700 → 13,000; 12,400 → 12,000; 15,500 → 16,000; 14,500 → 14,000 — are boundary tests in this calculator's suite, along with the values immediately either side of each tie.

A worked example.

Frequently asked questions.

References& sources.

1. [1]US Food and Drug Administration, Bacteriological Analytical Manual, Chapter 3 'Aerobic Plate Count', January 2001 Edition (original source BAM 8th ed. Rev. A, 1998; L. Maturin and J. T. Peeler). Introduction: 'The suitable colony counting range (10) is 25-250.' §B: decimal dilutions by transferring 10 mL of the previous dilution into 90 mL of diluent, 1 mL plated in duplicate. §C: EAPC, TNTC and no-colony reporting rules. §D: 'report only the first two significant digits', the round-half-to-even tie rule with four worked examples, and the weighted-mean formula N = Σc/[(n₁+0.1n₂)d] with the 537/0.022 = 24,409 ≈ 24,000 example this calculator reproduces. Read directly from the FDA PDF, 2026-07-29. Free.
2. [2]ISO/TC 34/SC 9 (Food products — Microbiology), 'Excel tool to implement the calculations of the colony-count technique according to ISO 7218 — Verification Report', S. Grosz, 28 August 2020. A free ISO committee document reproducing the ISO 7218:2007/Amd 1:2013 §11.2.6 calculation examples: 168 + 14 at 10⁻²/10⁻³ → 1.7E+04; 168+215 and 14+25 → 1.9E+04; 53 at 10⁻² → 5.3E+03; and 0 colonies at 10⁻¹ → '<1,0E+01'. All four are permanent tests in this calculator. Read directly, 2026-07-29.
3. [3]ISO 7218:2007/Amd 1:2013, Microbiology of food and animal feeding stuffs — General requirements and guidance for microbiological examinations, §11 (enumeration methods, including the weighted-mean colony-count calculation). PAYWALLED — cited bibliographically. Its content is verified here only through the free ISO committee verification report above, which is precisely why the countable range on this page is an editable input rather than a hard-coded number.
4. [4]ISO 6887-1:2017, Microbiology of the food chain — Preparation of test samples, initial suspension and decimal dilutions for microbiological examination — Part 1: General rules. The governing standard for the decimal dilution series that feeds a plate count. PAYWALLED — cited bibliographically; scope confirmed from the ISO catalogue entry, retrieved 2026-07-29.
5. [5]World Health Organization, WHO Laboratory Manual for the Examination and Processing of Human Semen, §2.8.3: 'Report the average sperm concentration to two significant figures.' Cited as an independent statement of the same reporting rule from a completely different field, confirming that two significant figures is a general convention for counting-based concentration measurements rather than an FDA idiosyncrasy. Public PDF, retrieved 2026-07-29.