Bacterial Growth Calculator

Background.

A bacterial growth calculator projects a batch culture forward — or backward — through the three phases a flask actually goes through: a lag while the cells adapt, an exponential phase where the population doubles on a fixed clock, and a stationary plateau once the medium runs out. It answers three separate questions from the same four parameters. Given an inoculum density, a doubling time, a lag duration and an incubation time, it tells you where the culture will be. Given a target density instead of an incubation time, it tells you how long to wait. And given a target and a deadline, it tells you what density to seed tonight so the flask is where you need it in the morning — the back-dilution calculation that is probably the single most common reason anyone opens a growth model at all.

The model implemented here is the three-phase linear model of Buchanan, Whiting and Damert, in which the specific growth rate is exactly zero during lag and stationary phase and constant during the exponential phase. That is a deliberate simplification. Real lag-to-log transitions are smooth, and a Gompertz or Baranyi fit follows a measured curve more faithfully; Buchanan and colleagues compared all three against Escherichia coli O157:H7 data and found good agreement between their predicted curves, with the linear model proving the more robust of the three when experimental data are sparse. That is exactly the situation a calculator user is in — a couple of parameters, not a full curve — which is why it is the model used here.

Two scope limits matter enough to state before you read any number this page produces. First, this is not a food-safety tool. Predicting whether a food is safe requires temperature-dependent, matrix-specific models validated against the actual organism, of the kind maintained in ComBase and the USDA Pathogen Modeling Program; a constant doubling time typed into a box is not a substitute, and no output here should be used to decide whether something is safe to eat. Second, this is not an infection or clinical tool. It projects a population in a flask under conditions you supply, and says nothing about a pathogen in a host.

Everything the calculator returns is only as good as the parameters you feed it, and two of those parameters are measurements rather than constants. The doubling time and the lag are properties of your strain in your medium at your temperature with your aeration and your inoculum history, and nothing in the model predicts them from conditions. Escherichia coli K-12 MG1655 doubles roughly every 20 minutes in LB at 37 °C, which Sezonov, Joseleau-Petit and D'Ari measured directly during steady-state growth, but the same organism on a poor carbon source in minimal medium can take hours. If you do not have your own numbers, measure them: two optical-density readings inside log phase and a clock are enough.

The stationary-phase ceiling is the third value you must supply, and the default on this page is explicitly a placeholder rather than a recommendation. Where a culture saturates depends on medium, strain and aeration, and even for the best-studied case the published numbers spread widely: Sezonov and colleagues report LB cultures reaching about OD 7 at saturation, which converts to anywhere between roughly 1.2 and 5.6 billion cells per millilitre depending on which optical-density conversion factor you apply. Measure your own plateau and type it in. The calculator will clip its projection there and tell you the culture has entered stationary phase rather than pretending exponential growth continues.

One last caution about precision. Growth is exponential, so parameter error compounds exponentially. A ten percent error in the doubling time is only a ten percent error for one generation, but across six generations it becomes about a 1.5-fold error in the final density. Report two or three significant figures, not ten, and treat any projection more than a few generations ahead as an order-of-magnitude guide rather than a prediction.

What is bacterial growth calculator?

Bacterial growth in a closed (batch) culture follows a characteristic curve that Monod enumerated in 1949: a lag phase in which the cells adapt and the population does not increase, an exponential or logarithmic phase in which each cell divides on a fixed schedule so the population doubles at a constant interval, a deceleration into a stationary phase once a nutrient becomes limiting or a waste product accumulates, and eventually a death phase. The three-phase linear model keeps the first three of these and idealises them: growth rate zero during lag, constant during log, zero once the ceiling is reached. In that model the population at time t is N(t) = N₀ · 2^((t − λ)/t_d) for t greater than the lag time λ, capped at a carrying capacity K. The doubling time t_d, also called the generation time, is the interval a population needs to double; the lag time λ is conventionally defined not as the moment the first cell divides but as the x-axis intercept you get by extrapolating the exponential slope back to the starting density, which is the definition Zwietering and colleagues formalised and the one the standard curve-fitting packages report. The number of generations completed is log₂ of the fold increase, and the specific growth rate constant μ relates to the doubling time by μ = ln2 / t_d. Because the densities appear only as ratios, the model is indifferent to whether you count in CFU per millilitre, cells per millilitre, or total cells — but it is not safe to run it directly on optical density above the instrument's linear range, which for E. coli begins to fail near OD 0.8.

How to use this calculator.

1. 01Choose which of the three questions you are asking: final density, time to a target, or the inoculum to seed.
2. 02Enter your strain's doubling time in minutes. Measure it from two log-phase readings if you do not already know it.
3. 03Enter the lag time in minutes, using the extrapolated-intercept definition your fitting software reports. Enter 0 if you are back-diluting a log-phase culture into identical medium, which largely abolishes lag.
4. 04Replace the stationary-phase ceiling with your own measured plateau. The default is a placeholder, not a recommendation.
5. 05For final density: enter the inoculum density and the total incubation time, lag included.
6. 06For time to target: enter the inoculum density and the density you want to reach. The answer includes the lag.
7. 07For back-dilution: enter the target density and the time you have available. The answer is the density to seed.
8. 08Read the phase-at-the-end field before you trust the number — if it says stationary, the projection has been clipped at your ceiling and the true curve was bending well before that point.

The formula.

During the exponential phase every cell divides on the same clock, so the population multiplies by two every t_d. After n doublings the population is N₀ · 2ⁿ, and n is simply the growth time divided by the doubling time. The lag phase is handled by subtracting it from the clock rather than by changing the rate: growth time is t − λ, floored at zero so that a culture still inside its lag returns its inoculum unchanged rather than a negative exponent. This is precisely how Hall, Acar, Nandipati and Barlow treat lag in the GrowthRates program — the lag is handled as though no growth occurred until the end of the lag phase, at which point growth becomes instantly exponential. Putting the two together gives N(t) = N₀ · 2^((t − λ)/t_d) for t greater than λ.

The stationary phase is a hard ceiling. If the exponential projection would carry the culture past the carrying capacity K you entered, the calculator returns K instead, recomputes the generation count as log₂(K/N₀), and reports the time at which the culture actually reached the ceiling rather than the full incubation. This is the third phase of the Buchanan three-phase linear model, in which the growth rate returns to zero. It is a clip, not a smooth logistic approach, so near the ceiling the real curve will already be bending while this model is still climbing at full rate — treat any result labelled stationary as an upper bound.

The same relation is written equivalently in natural logarithms. With μ = ln2 / t_d as the specific growth rate constant, the exponential segment is ln(N/N₀) = μ(t − λ), which is the form Zwietering and colleagues and the predictive-microbiology literature use. Both routes give identical numbers: for a 30-minute doubling time, μ = 0.6931471806 ÷ 0.5 h = 1.3862943611 h⁻¹, and over three hours of growth ln(N/N₀) = 4.1588830834, so N/N₀ = 64 — the same 2⁶ the base-2 route gives.

Inverting the relation gives the other two modes. Solving for time, the generations required are log₂(N_target/N₀), so the total time is λ + log₂(N_target/N₀) · t_d. Solving for the inoculum, the fold increase over the available window is fixed at 2^((T − λ)/t_d), so the seeding density is simply the target divided by that factor. All three modes are algebraic rearrangements of one equation, which is why round-tripping any two of them returns you to where you started.

Dimensionally, the densities cancel: N/N₀ and K/N₀ are ratios of quantities in the same unit, so their logarithms are pure numbers and the generation count is dimensionless. The doubling time and the lag are entered in minutes and the incubation in hours, and the calculator converts between them explicitly — a 90-minute lag removes exactly 0.5 hours from a 4-hour incubation. Rounding happens once, at the moment the results are returned; every threshold comparison, including whether the culture has passed its lag and whether it has reached its ceiling, is made at full precision so that a value a fraction below the ceiling is treated as below it rather than clipped.

The model refuses several inputs rather than returning a misleading number. A zero doubling time would divide by zero. A negative lag is not defined. An inoculum at or above the ceiling has nothing to grow into. A target above the ceiling can never be reached, so there is no time to return. And a target at or below the inoculum is rejected in the time mode, because this model has no death phase and a negative growth time is not an answer.

A worked example.

Frequently asked questions.

References& sources.

1. [1]Buchanan RL, Whiting RC, Damert WC (1997). "When is simple good enough: a comparison of the Gompertz, Baranyi, and three-phase linear models for fitting bacterial growth curves." Food Microbiology 14(4):313–326. The three-phase linear model implemented here: specific growth rate zero in lag and stationary phases, constant in the exponential phase; compared against E. coli O157:H7 data. Publisher full text is PAYWALLED; abstract and model statement verified 2026-07-29.
2. [2]Hall BG, Acar H, Nandipati A, Barlow M (2014). "Growth Rates Made Easy." Molecular Biology and Evolution 31(1):232–238. Lag handling: the program "treats the lag as though no growth occurred until the end of the lag phase at which point growth becomes instantly exponential"; model dN/dt = αN, N = N₀e^(αt). Retrieved 2026-07-29.
3. [3]Zwietering MH, Jongenburger I, Rombouts FM, van 't Riet K (1990). "Modeling of the Bacterial Growth Curve." Applied and Environmental Microbiology 56(6):1875–1881. Independent second authority on the lag convention: λ is defined as the x-axis intercept of the tangent to ln(N/N₀) versus time. Retrieved 2026-07-29.
4. [4]Sezonov G, Joseleau-Petit D, D'Ari R (2007). "Escherichia coli physiology in Luria-Bertani broth." Journal of Bacteriology 189(23):8746–8749. Independent empirical anchor: K-12 MG1655, "The doubling time during steady-state growth in Luria-Bertani broth was 20 min"; saturation near OD₆₀₀ ≈ 7. Retrieved 2026-07-29.
5. [5]Myers JA, Curtis BS, Curtis WR (2013). "Improving accuracy of cell and chromophore concentration measurements using optical density." BMC Biophysics 6:4. Optical density deviates from linearity for an E. coli sample at OD ≈ 0.8; "dilution to below OD≈0.5 is often recommended". Retrieved 2026-07-29.
6. [6]Monod J (1949). "The Growth of Bacterial Cultures." Annual Review of Microbiology 3:371–394. Source of the canonical growth-phase nomenclature (lag, acceleration, exponential, retardation, stationary, decline). BIBLIOGRAPHIC / PRINT-ARCHIVAL reference — the scanned PDF is not machine-readable, and no numeric claim on this page depends on it.