Log Reduction Calculator

Background.

A log reduction calculator converts between three ways of saying the same thing: how many powers of ten a microbial population fell by, what percentage of it was removed, and how many organisms are left. Enter the counts before and after a treatment and it returns the log reduction; enter a target log reduction and it returns the survivors; enter a percent-kill claim from a label or a specification and it tells you how many logs that actually is.

The relationship is short. Log reduction is log₁₀(N₀ ÷ N), where N₀ is the population before treatment and N the population after. Because it is a ratio, the unit cancels — colony-forming units per millilitre, plaque-forming units per millilitre, cells per millilitre or an absolute total all work, provided you use the same unit for both counts. Percent reduction is (1 − N ÷ N₀) × 100, which means 1 log is 90%, 2 logs is 99%, 3 logs is 99.9%, and every further log adds another nine.

That mapping is not a convention someone chose; it is arithmetic, and both US regulators that use it say so explicitly in their own words. EPA's Surface Water Treatment Rule at 40 CFR 141.70(a) requires water systems to achieve 'at least 99.9 percent (3-log) removal and/or inactivation of Giardia lamblia cysts' and 'at least 99.99 percent (4-log) removal and/or inactivation of viruses' — writing both forms in the same sentence. FDA's juice HACCP rule at 21 CFR 120.24(a) requires 'a 5 log (i.e., 10⁵) reduction … in the pertinent microorganism', pinning logs to fold-reduction. Two independent agencies, two different media, and their phrasings compose exactly. This calculator names which of those levels a given result reaches.

There is a reason to think in logs rather than percentages, and it shows up as soon as you look at the surviving fraction. Going from 99% to 99.9% sounds like a marginal improvement of nine tenths of a percentage point; in log terms it is a whole extra log, a tenfold reduction in survivors. Going from 99.999% to 99.9999% sounds like almost nothing and is again a full log. On a starting load of 10⁸ organisms per millilitre, a '99.99% effective' disinfectant still leaves 10⁴ per millilitre. The calculator reports the surviving fraction alongside the percent reduction for exactly this reason — the two framings feel very different and only one of them scales with the starting load.

Two things this calculator deliberately will not do. It will not accept a surviving count of zero, and it will not accept a percent reduction of exactly 100. Both imply an infinite log reduction, and no plate, assay or measurement can demonstrate that a population is truly zero — a blank plate only tells you the survivors are below the detection limit of the method. The correct report in that case is a 'greater than' value computed by substituting the detection limit for the survivor count, and the CFU calculator on this site computes that limit for you. Claiming a specific log reduction from a blank plate over-states what the experiment actually showed, and it is the most common way a log-reduction figure becomes indefensible.

It also has no D-value or decimal-reduction-time mode, and that omission is deliberate rather than an oversight. D = t ÷ log reduction is undefined when the log reduction is zero — and 'the treatment achieved nothing' is a real result this page must be able to express. A D-value also presumes log-linear inactivation kinetics, and real survivor curves routinely show shoulders and tails that two data points cannot detect. Reporting a D-value from a single before-and-after pair asserts a kinetic model the data cannot support.

Finally: a log reduction is only defined for the conditions it was measured under. The same disinfectant at the same concentration gives different log reductions against different organisms, at different contact times and temperatures, at different pH, and above all in the presence of organic soil. A number without its conditions is not transferable, and negative results are permitted and reported here — if the count after treatment is higher than before, the calculator says the population grew rather than quietly clamping the answer at zero.

What is log reduction calculator?

A log reduction is the base-10 logarithm of the ratio between a microbial population before and after a treatment. It is the standard currency of disinfection, sterilisation, pasteurisation and water treatment because microbial inactivation is multiplicative rather than additive: a process that removes 90% of what is present removes 90% whether it starts with a thousand organisms or a billion, so the natural scale is logarithmic.

One log reduction is a tenfold drop, or 90% removal. Two logs is a hundredfold, 99%. Three is 99.9%, four 99.99%, five 99.999%, six 99.9999%. Each additional log removes 90% of whatever survived the previous one, which is why the percentages accumulate nines and why the differences between them look small in percentage terms while being large in practice.

Regulators specify performance in logs precisely because the scale is load-independent. EPA's Surface Water Treatment Rule requires 3-log (99.9%) removal or inactivation of Giardia lamblia cysts and 4-log (99.99%) of viruses between the raw-water intake and the first customer. FDA's juice HACCP rule requires processors to include controls achieving at least a 5-log (10⁵) reduction in the 'pertinent microorganism', which the regulation defines as 'the most resistant microorganism of public health significance that is likely to occur in the juice'. Sterilisation goes further and works in sterility assurance levels, expressing the probability that a single unit remains non-sterile.

One scope difference between those two rules is worth noticing rather than glossing over. EPA's requirement is for removal *and/or inactivation* — physical removal by filtration earns log credit alongside chemical or UV inactivation. FDA's juice rule is a reduction in the pertinent microorganism achieved by process controls. The arithmetic is identical in both cases; what counts as having achieved it is not. That is why a log reduction number, on its own, is never compliance: the regulations require a validated process, and the number is one piece of evidence about it.

How to use this calculator.

1. 01Measure the population before treatment and after it, using the same method on comparable samples — usually a plate count, which the CFU calculator converts into CFU/mL for you.
2. 02Choose 'log reduction achieved' and enter both counts. Any unit works as long as it is the same for both, because the ratio cancels it.
3. 03If nothing grew on the post-treatment plate, do NOT enter zero. Substitute your method's detection limit (1 ÷ the volume of original sample plated) and read the answer as a 'greater than' value — that is what the experiment actually demonstrated.
4. 04Read the surviving fraction as well as the percent reduction. On a heavy starting load, a very high percentage can still leave a large absolute number of organisms.
5. 05To plan rather than measure, switch to 'survivors remaining' and enter your starting load and the log reduction your standard requires — 3 or 4 for EPA drinking water, 5 for FDA juice HACCP.
6. 06To decode a marketing or specification claim, switch to 'from a percent kill' and type the claimed percentage. This is the fastest way to see that '99.9% effective' is 3 logs and '99.99%' is 4.
7. 07Record the conditions with the number: organism, starting concentration, contact time, temperature, pH and soil load. A log reduction measured under one set of conditions does not transfer to another.
8. 08Report log reductions to one or two decimal places. The counts underneath rarely support more — plate counts themselves are reported to two significant figures.

The formula.

Everything on this page follows from one ratio. The reduction factor is N₀ ÷ N, and the log reduction is its base-10 logarithm, log₁₀(N₀ ÷ N). Because both counts appear only as a ratio, the units cancel and the result is dimensionless — which is why the same number describes a treatment whether the counts were CFU/mL, PFU/mL, or absolute totals, and why scaling both counts by any factor leaves the log reduction unchanged.

Working the example: a starting titre of 1.0 × 10⁶ CFU/mL and a post-treatment titre of 1.0 × 10² CFU/mL give a reduction factor of 10⁶ ÷ 10² = 10,000, so the log reduction is log₁₀(10,000) = 4. The surviving fraction is 100 ÷ 10⁶ = 10⁻⁴, which is 0.01%, so the percent reduction is 100 − 0.01 = 99.99%. Both the 3-log Giardia level and the 4-log virus level in EPA's 40 CFR 141.70(a) are numerically reached; the 5-log level in FDA's 21 CFR 120.24(a) is not.

The two inverse forms come from rearranging the same equation. Given a target log reduction, the survivors are N = N₀ ÷ 10^LR — so a 5-log reduction on 10⁶ organisms leaves 10⁶ ÷ 10⁵ = 10, and the percent reduction is 99.999%. Given a claimed percent reduction p, the surviving fraction is 1 − p ÷ 100 and the log reduction is −log₁₀(1 − p ÷ 100) — so 99.9% is −log₁₀(0.001) = 3 logs exactly. The three modes are one relation solved for three different unknowns, and the test suite asserts that each round-trips back to the others.

Negative log reductions are permitted and reported. If the count after treatment exceeds the count before, the ratio is less than 1 and the logarithm is negative, meaning the population grew — a real outcome for an under-dosed treatment, a failed neutralisation step, contamination during sampling, or growth between sampling and plating. Clamping that at zero would hide the most diagnostically useful result the experiment can produce.

Rounding stage: every value is computed at full working precision and rounded only at the return boundary, to twelve significant figures. One rule is specific to this page and worth stating: **the value used to decide which regulatory level a result reaches is the same rounded value that is displayed.** The log reduction is rounded exactly once, and both the number you see and every 'reaches the 3-log level' comparison read that single value. That makes it structurally impossible for the page to display 3.00 while classifying the result as below 3 — the classic boundary defect that appears only at a threshold and never in the middle of a range. Boundary tests sit immediately before, at, and immediately after 3, 4 and 5 logs, and on every percent-to-log pair from 90% to 99.9999%.

A worked example.

Frequently asked questions.

References& sources.

1. [1]US Environmental Protection Agency, 40 CFR 141.70(a) — National Primary Drinking Water Regulations, Subpart H (Filtration and Disinfection), General requirements. Verbatim: 'at least 99.9 percent (3-log) removal and/or inactivation of Giardia lamblia cysts' and 'at least 99.99 percent (4-log) removal and/or inactivation of viruses between a point where the raw water is not subject to recontamination by surface water runoff and a point downstream before or at the first customer.' This is the regulation that states the percent↔log mapping explicitly in both forms. Retrieved 2026-07-29.
2. [2]US Food and Drug Administration, 21 CFR 120.24(a) — Hazard Analysis and Critical Control Point (HACCP) Systems, Subpart B (Pathogen Reduction), Process controls. Verbatim: control measures achieving 'at a minimum, a 5 log (i.e., 10⁵) reduction, for a period at least as long as the shelf life of the product when stored under normal and moderate abuse conditions, in the pertinent microorganism', with 'pertinent microorganism' defined as 'the most resistant microorganism of public health significance that is likely to occur in the juice'. Independent second authority: a different agency, a different medium, and its log↔fold phrasing composes exactly with EPA's percent↔log phrasing. Retrieved 2026-07-29.
3. [3]US Food and Drug Administration, Bacteriological Analytical Manual, Chapter 3 'Aerobic Plate Count', January 2001 Edition. §C.4: when plates from all dilutions have no colonies, 'report APC as less than 1 times the corresponding lowest dilution used' — the rule that governs the denominator of a log reduction when the post-treatment plate comes back blank. §D: the two-significant-figure reporting rule for the counts a log reduction is built from. Free, read directly 2026-07-29.
4. [4]ISO/TC 34/SC 9, 'Excel tool to implement the calculations of the colony-count technique according to ISO 7218 — Verification Report', S. Grosz, 28 August 2020. Documents the '<1,0E+01' detection-limit convention for a plate with no colonies, from a standards body independent of the FDA — the same convention this page applies when refusing a survivor count of zero. Free ISO committee document, retrieved 2026-07-29.