Time-Kill Assays: Measuring Bactericidal Kinetics Over Time

A minimum inhibitory concentration (MIC) tells you the lowest drug concentration that prevents visible growth after a fixed incubation period — usually 18 to 24 hours. It is a single endpoint, read once, and it cannot distinguish a drug that has killed the inoculum from one that has simply stopped it from multiplying. A broth microdilution MIC and a bactericidal drug can produce identical-looking wells. The time-kill assay is the method that resolves that ambiguity: it tracks viable colony counts over time against a fixed drug exposure, producing a kinetic curve instead of a static number.

What a Time-Kill Assay Measures That MIC Cannot

MIC and time-kill answer different questions. MIC answers “does growth resume at this concentration by the endpoint?” — a yes/no threshold. A time-kill assay answers “how much of the starting population is actually dead, and how fast?” by sampling the same culture repeatedly and plating for viable colony-forming units (CFU/mL) at each timepoint. The result is a curve of log₁₀ CFU/mL against time, not a single cutoff.

This distinction matters clinically because MIC alone cannot separate a bactericidal agent (one that actively reduces the viable population) from a bacteriostatic one (one that only arrests growth, leaving clearance to host immune defenses). In an immunocompetent patient with a routine infection, that distinction is often academic. In endocarditis, meningitis, osteomyelitis, febrile neutropenia, or any site where host clearance is limited or the organism is hard to reach, it is not — bactericidal activity, and how quickly it’s achieved, becomes a real treatment-selection variable.

Assay Setup: Sampling Viable Counts Against a Fixed Exposure

The core design is straightforward, though execution detail (media, inoculum density, sampling frequency) needs to be tight enough to make the resulting curve reproducible:

• Standardized starting inoculum. A log-phase bacterial suspension is adjusted to a defined starting density — conventionally in the range of 5 × 10⁵ to 1 × 10⁶ CFU/mL — the same order of magnitude used for standard broth microdilution susceptibility testing.
• Drug exposure. The inoculum is exposed to either a single fixed concentration (commonly a clinically achievable multiple of the MIC, such as 1×, 4×, or 8× MIC) or, more informatively, a concentration series tested in parallel — plus an antibiotic-free growth control to confirm the inoculum itself remained viable and grew as expected.
• Serial sampling. Aliquots are withdrawn at defined intervals — typically 0, 1, 2, 4, 6, 8, and 24 hours, sometimes extended further for slow-killing agents — and immediately serially diluted (see the site’s serial dilution technique for the underlying calculation) to bring counts into a plateable range.
• Drug carryover control. Because residual antibiotic in a sampled aliquot can suppress colony formation on the plate independent of actual kill, dilution past the drug’s effective concentration, a wash step, or a neutralizing agent is used before plating — otherwise the curve overstates killing.
• Plating and counting. Each dilution is plated on drug-free agar, incubated, and colonies counted to back-calculate CFU/mL at that timepoint. Plotting log₁₀ CFU/mL against time produces the time-kill curve.

The Bactericidal Threshold: a ≥3-log₁₀ Reduction

The working definition used in CLSI methodology for bactericidal activity (CLSI M26, the standard governing bactericidal-activity testing methods) is a reduction of at least 3 log₁₀ CFU/mL — a 99.9% drop — from the starting inoculum, generally assessed by 24 hours unless the curve shows the threshold is reached earlier. Anything short of that — growth suppressed relative to the untreated control, but without a 3-log drop — is read as bacteriostatic rather than bactericidal, even if the well would still read “susceptible” on an MIC panel.

This is the kinetic analog of the minimum bactericidal concentration (MBC), which asks the same bactericidal question but as a single static endpoint — the lowest concentration that produces ≥99.9% kill by a fixed timepoint, read the same way an MIC is read. A time-kill curve subsumes that information and adds the dimension MBC alone omits: how fast killing happens, and whether it holds up over the full dosing interval rather than just at one arbitrary read time. A curve that shows an early 3-log drop followed by regrowth by 24 hours — a real, documented pattern with some organism-drug combinations, often reflecting adaptive resistance or a subpopulation with reduced susceptibility — is a materially different (and clinically worse) result than a curve that reaches and holds the same 3-log threshold, even though a single 24-hour CFU count might look similar to a same-day MBC read.

Concentration-Dependent vs. Time-Dependent Killing

Running time-kill curves across a concentration series, rather than at one fixed multiple of MIC, is what reveals a pharmacodynamic pattern that a single MIC value structurally cannot show: whether the rate and extent of killing keep increasing as concentration rises, or whether killing saturates once concentration clears a modest multiple of MIC.

• Concentration-dependent killing. The curve keeps dropping faster and further as concentration increases well beyond the MIC, with no clear plateau across the tested range. Aminoglycosides and fluoroquinolones characteristically show this pattern. The pharmacodynamic driver is peak exposure relative to MIC (C_max/MIC) or total exposure relative to MIC (AUC/MIC), not sustained exposure.
• Time-dependent killing. The curve flattens once concentration exceeds roughly 4–5× the MIC — pushing the concentration higher doesn’t meaningfully accelerate or deepen the kill. Beta-lactams characteristically show this pattern. The pharmacodynamic driver is the proportion of the dosing interval that free drug concentration stays above the MIC (%T>MIC), not how high the peak goes.

Neither pattern is visible from an MIC value alone, because MIC is read at one concentration and one timepoint by design. The pattern only emerges once killing is tracked as a curve across both dimensions — time and concentration — which is the entire premise of the assay.

Why This Distinction Drives Dosing-Regimen Design

The concentration-dependent/time-dependent classification is not just descriptive — it is the basis for how dosing intervals and infusion strategies get chosen once a drug class’s killing pattern is established:

• For concentration-dependent agents, the dosing goal is to maximize the peak-to-MIC ratio, which is the pharmacodynamic logic behind once-daily (extended-interval) aminoglycoside dosing rather than smaller, more frequent doses — a higher single peak drives more killing than the same total daily dose spread thinner.
• For time-dependent agents, the dosing goal is to maximize the fraction of the interval spent above MIC, which is the logic behind extended or continuous infusion of beta-lactams, or shortening the interval between doses, rather than pushing the peak higher — a taller peak that clears quickly buys little over a lower, sustained exposure.

Time-kill data is where this classification is first established in vitro, before it is carried forward into animal infection models and ultimately clinical pharmacokinetic-pharmacodynamic targets. A susceptibility report built only from MIC values — the kind an antibiogram aggregates at the population level — tells a clinician whether an organism is susceptible at achievable concentrations; it does not tell them whether the dosing interval or the infusion strategy actually matters for that drug-organism pair. That second question is answered by the killing pattern the time-kill curve exposes, and it is a routine input into antimicrobial stewardship dosing-optimization work, particularly for extended-infusion beta-lactam protocols in critically ill or difficult-to-treat infections.

Practical Considerations and Common Pitfalls

• Inoculum effect. Some organism-drug combinations show reduced killing at higher starting inocula than the standard range, which is part of why the starting density is standardized rather than left to vary — an unusually high inoculum can make a genuinely bactericidal combination look bacteriostatic.
• Regrowth after initial kill. A curve that drops 3-log by 6 hours but rebounds toward the growth control by 24 hours is a distinct, clinically important finding — often attributable to a resistant subpopulation being selected under drug pressure — that a single-timepoint MBC read would miss entirely if that read happened to fall before the regrowth.
• Distinguishing from synergy testing. A time-kill design extended to combinations of two drugs (checkerboard-style concentration pairs, or a fixed-ratio combination tracked over time) is used to test for synergistic, additive, or antagonistic killing between agents — a related but separate application from the single-agent bactericidal-kinetics question this page describes.
• Carryover artifact. As noted above, failing to dilute or neutralize residual drug before plating is one of the more common sources of an artificially steep-looking kill curve — worth checking before accepting an unusually dramatic result.

Frequently Asked Questions

Is a time-kill assay the same as a minimum bactericidal concentration (MBC) test?

No. MBC is a single-timepoint endpoint — the lowest concentration producing ≥99.9% (3-log₁₀) kill by a fixed read time, determined the same way an MIC is determined. A time-kill assay tracks the same ≥3-log₁₀ bactericidal threshold but as a curve across multiple timepoints, which is the only way to see the rate of killing, whether the threshold is sustained, or whether regrowth occurs after an initial drop.

Can a drug be bactericidal against one organism and bacteriostatic against another?

Yes. Bactericidal versus bacteriostatic is not a fixed property of a drug in isolation — it depends on the drug-organism combination, and sometimes the concentration tested. A drug that reliably reaches a 3-log₁₀ kill against one species at achievable concentrations may only suppress growth against another, which is exactly why time-kill data is generated per organism rather than assumed from drug class alone.

What starting inoculum and sampling schedule does time-kill methodology use?

CLSI methodology for bactericidal-activity testing (CLSI M26) is built around a standardized starting inoculum in the same order of magnitude used for broth microdilution — roughly 5 × 10⁵ to 1 × 10⁶ CFU/mL — with viable counts sampled at multiple timepoints (commonly 0, 1, 2, 4, 6, 8, and 24 hours) rather than read once.

Why does this matter if the organism already tests “susceptible” on an MIC panel?

Susceptible/resistant categorization from an MIC tells you the drug can suppress growth at achievable concentrations — it says nothing about how fast or how completely the organism is actually killed, or which dosing strategy (higher peak vs. longer time above MIC) will make the drug perform better in vivo. Those are exactly the questions a single MIC endpoint is structurally unable to answer.

Does a time-kill curve replace clinical dosing guidance?

No — it is upstream, mechanistic evidence. Time-kill kinetics establish the concentration-dependent versus time-dependent classification for a drug class in vitro; that classification is then combined with pharmacokinetic data and clinical outcome studies to set actual dosing recommendations. A time-kill curve does not itself specify a dose or interval for a patient.

Time-kill methodology sits alongside MIC, MBC, and antibiogram data as one more layer of evidence for how an antimicrobial actually behaves against a given organism — not a replacement for any of them, but the one that adds the time dimension none of the others capture. For the broader trial-operations and lab-methodology context this fits into, see the clinical research cluster.