How to count bacterial colonies: the 30-300 rule, TNTC and CFU per mL

Which countable range belongs to which method

The countable range comes from the method you are following, not from microbiology in general, and every version of it is colonies per plate, never CFU per mL. FDA's Bacteriological Analytical Manual, Chapter 3 (BAM) puts the main ones side by side for the pour plate: 30-300 by USDA and AOAC, 25-250 by the American Public Health Association (APHA), and 15-300 by ISO. Spread plates, track plates and membrane filters add their own.

The shorthand that circulates in lab handouts, "pour 25-250, spread 30-300, membrane 20-200", is a mix of these. The 25-250 pour range was BAM's until 2025 and is still APHA's. 30-300 is used for pour plates as much as spread plates. 20-200 is ASM's window for track plates and is widely repeated for membrane filters, but the membrane method we could open, EPA Method 1604, counts filters with no more than 200 colonies in all and prefers 20-80 of the target colonies (clause 12.1). If your lab's procedure names a standard, its range wins. In a teaching lab with no standard named, 30-300 is the convention the American Society for Microbiology teaches.

Where 30-300, 25-250 and 15-300 come from

All three are compromises between two kinds of error, and the first measurements were made on milk. In 1916, in the first year of the Journal of Bacteriology, R. S. Breed and W. D. Dotterrer found plates with more than 40 and fewer than 200 colonies the most reproducible, and advised disregarding plates with fewer than 20 or more than 400, as George O'Toole's 2016 retrospective in the same journal recounts. O'Toole notes that most manuals now give "somewhere between 30 and 300 colonies (or some similar range)", and does not trace where that figure came from.

The 25-250 range has a clear source. In 1980 Tomasiewicz and colleagues, in the Journal of Food Protection, combined bias and variance from triplicate plates of serially diluted raw milk and found that the best range depends on the dilution step: 70-140 colonies for twofold dilutions, 40-200 for fivefold and 25-250 for tenfold (abstract on FAO AGRIS). 25-250 is therefore a range for tenfold series in particular.

The 15-300 range is the newest. FDA's March 2025 revision of BAM Chapter 3 changed its range from 25-250 to 15-300 per plate, giving two reasons: ISO is internationally accepted in regulatory work, and most published studies report plates with low counts. The January 2026 edition keeps 15-300 and calculates with a maximum-likelihood method, set out below.

Pour plate, spread plate and membrane filter

The plating method changes how much sample a plate holds and how soon it crowds, which is why the ranges differ. British Columbia's heterotrophic plate count method, which follows Standard Methods 9215, sets out the trade-offs.

• Pour plate. The sample is mixed into molten agar. In BAM, 1 mL of each dilution goes into duplicate dishes with 12-15 mL of plate count agar cooled to 45 ± 1 °C, incubated 48 ± 2 h at 35 °C. A pour plate takes 0.1 to 2.0 mL, and its colonies are small and compact, less likely to run into each other than surface colonies.
• Spread plate. A small volume is spread over the surface of a pre-dried plate. ASM's spread plate protocol calls 0.1 mL the convenient volume, and the BC method puts what the agar can absorb at 0.1 to 0.5 mL. There is no heat shock, and every colony is on the surface, easy to tell from a bubble or a particle. Because only 0.1 mL is plated, an undiluted sample needs at least 300 CFU per mL to reach 30 colonies.
• Membrane filter. A large volume of clear water is drawn through a 47 mm filter, which then sits on the agar. It suits low numbers in large volumes, but the filter's display area is small: about 17 cm² against about 64 cm² for a 90 mm dish, from the nominal diameters.

The dish size changes the ceiling. ISO 7218's amendment of 2001 fixes its limits for 90 mm dishes and says that for other diameters "the maximum number of colonies shall be increased proportionately to the surface area". A 150 mm dish has (150 ÷ 90)² ≈ 2.8 times the area, so its ceiling is about 830. BAM's spiral plate method, on 100 or 150 mm dishes, uses a different rule again: count from the rim inwards until 20 colonies have been counted, then divide by the volume in those segments.

The CFU per mL formula

The formula is CFU per mL = colonies ÷ (dilution × volume plated in mL), the "standard formula" in ASM's serial dilution protocol. The dilution is written as a fraction, 10⁻⁵ rather than 100,000, and the volume is what went onto the plate, usually 1 mL for a pour plate and 0.1 mL for a spread plate. The two multiplied together are the amount of the original sample on the plate; BAM's 2026 edition calls it V and gives 1:100 × 1 mL = 0.01 mL as its example. For a food, the first dilution is a homogenate, so the result is per gram.

When more than one plate is countable, food methods pool them. ISO 7218 writes it as N = ΣC ÷ [V × (n₁ + 0.1 n₂) × d], where ΣC is every colony on the retained plates, n₁ and n₂ the number of plates at the first and second dilution, d the first dilution and V the volume plated. That is simply all the colonies divided by all the original sample they came from. BAM's 2026 edition shows the same thing is the maximum-likelihood estimate, a weighted average in which each plate counts in proportion to the sample it received, whenever no plate is TNTC.

Results are reported to two significant figures, rounded only at the end. The rounding rule differs: BAM rounds a third digit of 5 towards an even second digit, so 14,500 becomes 14,000 and 15,500 becomes 16,000, while ISO 7218 and the Standard Methods procedure below round any 5 up.

Worked example 1: a spread plate series

A culture is diluted tenfold to 10⁻⁶, and 0.1 mL of the last three dilutions is spread on duplicate plates. After incubation: 10⁻⁴, both plates over 300; 10⁻⁵, 152 and 168; 10⁻⁶, 12 and 9.

1. Choose the plates. Only 10⁻⁵ is in range under every rulebook above. The 10⁻⁴ plates are TNTC, and 12 and 9 are below 15, 25 and 30 alike.
2. Average the duplicates. (152 + 168) ÷ 2 = 160.
3. Divide by the original sample on the plate. 10⁻⁵ × 0.1 mL = 10⁻⁶ mL, so 160 ÷ 10⁻⁶ = 160,000,000.
4. Report it. 1.6 × 10⁸ CFU per mL.

The 10⁻⁶ plates show why the floor exists. A perfect tenfold step from 160 predicts 16; the plates gave 12 and 9, which on their own would have given about 1.1 × 10⁸, a third lower.

Worked example 2: two countable dilutions on pour plates

This is the worked example from BAM's 2001 edition, with 1 mL poured from each dilution in duplicate: 232 and 244 colonies at 10⁻², 33 and 28 at 10⁻³. All four are inside 25-250, BAM's range at the time, and inside 15-300, its range now.

1. Add every retained colony. 232 + 244 + 33 + 28 = 537.
2. Add the original sample those plates received. Two plates at 0.01 mL and two at 0.001 mL make 0.022 mL, the same as (2 + 0.1 × 2) × 10⁻² in the ISO formula.
3. Divide. 537 ÷ 0.022 = 24,409.
4. Round. 24,000 CFU per mL, or per gram for a food: 2.4 × 10⁴.

A 30-300 rulebook drops the plate with 28, giving 509 ÷ 0.021 = 24,238, which still rounds to 24,000. The answer only moves when the plates are combined a different way. Water methods following Standard Methods 9215 B convert each dilution to CFU per mL first and average the results; Georgia EPD's procedure (SOP 5-002, clause 11.2.1) shows 280 colonies from 0.01 mL and 34 from 0.001 mL as 28,000 and 34,000, reported as 31,000 CFU per mL.

Pooling gives most weight to the plate with more colonies, which is the more precise one; averaging gives the 34-colony plate an equal say. Neither is an error. Report by the method your procedure cites, and say which one it is.

TNTC, too few to count and zero

TNTC means too numerous to count: more colonies than the ceiling, or growth merged into a lawn. It is a label, not a number, and it is never averaged in as 300. Below the floor, teaching labs write TFTC, too few to count; the official methods keep those counts and report them as estimates.

Three worked figures from the sources make the rules concrete. ISO 7218's amendment counts 12 and 13 colonies at 10⁻² as an estimated 25 ÷ 0.02 = 1,250, reported as 1,300. The Georgia procedure reports 22 colonies from 0.1 mL as 220 CFU per mL, estimated value. BAM's 2001 edition handled plates with more than 100 colonies per cm² as greater than 100 × the highest dilution × the plate area: on a 65 cm² plate at 1:1000, greater than 6,500,000.

BAM's 2026 edition replaced those estimates with arithmetic that keeps the TNTC plates: each one enters as a lower bound, the visible count if there is one or a threshold such as 300, and the estimate is computed in software: the MPN package for R that BAM cites, and its apc function. A spreader is counted as one source: BAM counts a chain of colonies from one broken-up clump as one colony, which is also the plainest reason a colony-forming unit is not a cell.

Why the limits exist: counting error

The floor exists because small counts scatter. BAM's 2026 edition assumes colony counts follow a Poisson distribution, in which the standard deviation of a count is its square root, so the error relative to the count is 1 ÷ √N. At 15 colonies that is about 26%; at 300 it is under 6%.

The ceiling exists for a different reason. BAM's 2001 edition warns that crowded plates "may be difficult to count or may inhibit the growth of some bacteria, resulting in a low count", and ASM adds that high counts are confounded by small colonies and overlapping ones. At the low end, ASM notes that air contaminants weigh heavily on a small count. The Poisson figures cover counting only: pipetting and mixing add their own error, which is why BAM has every dilution shaken 25 times through a 30 cm arc within 7 seconds. The same square-root logic sets the precision of a manual white cell differential.

Counting a plate without losing your place

BAM's equipment list for the plate count includes a dark-field or Quebec colony counter, with a light and a grid plate, and a tally register. The counting itself is done the same way whatever counter you use:

1. Count every colony, pinpoint ones included, as BAM's 2001 counting guide says, on a plate held over the grid.
2. Mark each colony on the underside of the dish with a fine felt-tip dot as you count it, so none is counted twice and the plate can be rechecked.
3. Work square by square, in a fixed order, and finish one square before starting the next.
4. Record each plate with its dilution and volume before moving on, not the total for the dilution.

Simple Online Counter is a free, browser-based set of counters that keeps its counts in your browser's own storage. Its tally counter takes one tap per colony and has undo for a double tap. Its multi counter keeps one named counter per plate, "10⁻⁵ A" and "10⁻⁵ B", and exports them all as a CSV. Neither recognises colonies; the judgement about what is one colony remains yours, and the arithmetic above turns the counts into CFU per mL.

Mistakes the lab handouts repeat

Most of the pages that rank for this question are lab handouts, and they agree on the core method. Where they go wrong, the primary sources settle it:

What this guide does not cover

This guide covers counting plates and turning counts into CFU per mL for teaching, food and water laboratories. It is not a guide to clinical microbiology: clinical laboratories interpret colony counts from patient samples under their own validated procedures, and a count worked out here is not a diagnosis. Your laboratory's procedure, and the standard it cites, overrides every range on this page.

Two limits on the sources. ISO's standards are sold, not published: ISO 7218:2024 has replaced the 2007 edition, and the ISO clauses quoted here are from the 2001 amendment of the 1996 edition, so a laboratory working to ISO should check the current clause 11.2. Standard Methods 9215 is also subscription-only; its pour-plate rules are quoted from Georgia EPD's published procedure, and we could not check the 9215 D membrane-filter wording.