Counting bacterial colonies on an agar plate means tallying each visible, distinct colony as one “colony-forming unit” (CFU), then multiplying by the dilution factor to estimate the concentration of bacteria in the original sample. The process sounds straightforward, but getting an accurate count depends on having plates in the right colony range, using good technique to avoid double-counting, and understanding the quirks that can throw off your numbers. The details matter more than most people expect, especially when plates are crowded, colonies overlap, or the organism itself misbehaves.
Getting Plates Into a Countable Range
Before you count anything, you need plates with a manageable number of colonies. If you plate an undiluted sample of most bacterial cultures, you will get a lawn of confluent growth rather than individual colonies. That is why serial dilution is the standard first step: you dilute the original sample in a series of tenfold steps, plate multiple dilutions, incubate, and then select the plate with colonies you can actually distinguish and count individually.1Europe PMC. Maximum likelihood estimators for colony-forming units
The traditional countable range most labs use is 25 to 250 colonies per plate for the pour plate method, or 30 to 300 for the spread plate method. These thresholds exist because plates with fewer than about 25 colonies introduce too much statistical noise (a few colonies more or less changes your estimate dramatically), while plates with more than 250 or 300 start having so many overlapping colonies that accurate counting becomes impossible. If none of your dilutions land in this range, you either need to adjust your dilution series or note that the count is estimated rather than precise.
Once you have identified which plate falls within the countable range, the arithmetic is simple: multiply the colony count by the reciprocal of the dilution factor. If you counted 150 colonies on a plate from a 1:10,000 dilution, the estimated concentration is 150 × 10,000, or 1.5 million CFU per milliliter of the original sample.
How to Physically Count Colonies
The actual counting is more manual than people expect. In most teaching and clinical labs, someone sits at a bench with a colony counter, which is essentially a magnifying lens mounted over a light source with a grid background. You hold the plate against the illuminated grid, and each time you spot a colony, you mark the back of the plate with a fine-tipped marker so you do not count it twice. Some counters have a pressure-sensitive pen that adds to a digital tally each time you press it against the plate.
A few practical tips make this go more smoothly:
- Work systematically: Start at one edge of the plate and scan across in rows or work outward from the center in a spiral. Random scanning almost guarantees you will miss colonies or count some twice.
- Mark as you go: A dot with a felt-tip pen on the bottom of the plate, directly below each colony, is the most reliable way to keep track. Some people prefer to divide the plate into quadrants and count each one separately.
- Use oblique lighting: Tilting the plate slightly under the light makes translucent or pinpoint colonies easier to see. Colonies that are nearly invisible under direct overhead illumination can cast small shadows when lit from the side.
- Count every distinct colony as one CFU: Even if a colony is tiny compared to its neighbors, it still counts. A CFU represents whatever clump of cells gave rise to that colony, whether that was a single cell or a cluster.
Counting takes longer than you might think. One study evaluating manual versus automated counting found that experienced technicians spent an average of about 70 seconds per plate for manual counts, which adds up fast when you are processing dozens of plates in a session.2Europe PMC. Evaluation of an Automated System for the Counting of Microbial Colonies
Spread Plates, Pour Plates, and Spiral Plating
The method you used to plate your sample affects both what your colonies look like and how easy they are to count. The three main approaches each have trade-offs.
With a spread plate, you pipette a small volume onto the surface of solidified agar and spread it with a sterile spreader. All colonies grow on the surface, where they are easy to see and distinguish. With a pour plate, you mix the sample into molten agar before it solidifies, so colonies grow both on the surface and embedded within the agar. Embedded colonies tend to be smaller and can be harder to see. A comparison of the two methods for water samples found that the spread plate method was both more accurate and more precise than the pour plate for enumerating bacterial populations.3Journal AWWA. Standard plate count: A comparison of pour plate and spread plate methods
Spiral plating is a third option used in higher-throughput labs. A mechanical device deposits a decreasing concentration of sample in a spiral pattern across the plate surface, effectively building the dilution series into a single plate. You count colonies in defined arcs of the spiral and use a formula (or software) to calculate CFU. Spiral plating generally agrees well with conventional methods. One comparison using meat samples found that spiral plating and standard aerobic plate counts, whether counted manually or by laser scanner, fell within a half-log range of each other for most samples.4Journal of Rapid Methods & Automation in Microbiology. USE of SPIRAL PLATER and LASER COLONY SCANNER FOR ENUMERATION of MICROORGANISMS IN MEAT Spiral plating can also cut incubation time for slow-growing organisms: researchers working with Mycobacterium paratuberculosis found that spirally plated cultures could be microscopically counted after just 8 to 14 days instead of the typical 27 to 28 days needed for visible colonies, with no statistical difference in the final CFU count.5PubMed. The use of spiral plating and microscopic colony counting for the rapid quantitation of Mycobacterium paratuberculosis
When Colonies Do Not Cooperate
Counting assumes you have discrete, round, well-separated colonies. Real plates are often messier than that, and certain organisms make things particularly difficult.
Swarming is the classic headache. Species like Proteus mirabilis do not stay put. They produce thin films of motile cells that spread across the agar surface, sometimes covering the entire plate and obscuring other colonies. Proteus forms visible swarming terraces on standard agar concentrations between about 1% and 2.5%, and only forms isolated, countable colonies at higher agar concentrations around 4%.6bioRxiv. Swarming bacteria respond to increasing barriers to motility by increasing cell length and modifying colony structure If you are working with a sample that might contain swarming species, you can add anti-swarming agents to the medium. One effective option is p-nitrophenyl glycerin, which abolishes Proteus swarming for extended incubation periods without inhibiting the growth of other organisms like Staphylococcus or Streptococcus.7PubMed Central. Abolition of swarming of Proteus by p-nitrophenyl glycerin: application to blood agar media
Spreading and rhizoid colonies are a milder version of the same problem. Some organisms naturally produce colonies with irregular, branching edges that can merge with neighbors. When two spreading colonies touch, deciding whether to count them as one or two becomes a judgment call. Most guidelines say to count each distinct center of growth as one colony, even if their edges have merged.
Pinpoint colonies present the opposite challenge. Some bacteria, especially slow growers or nutritionally fastidious species, produce colonies so tiny that they are easy to overlook. This is where oblique lighting and a magnifying lens earn their keep. If pinpoint colonies are numerous and clearly a different morphology from the larger colonies on the same plate, note them separately in your report.
Lighting and Imaging Setup
Lighting is underrated in colony counting. Overhead fluorescent lab lights create glare on the agar surface and can wash out translucent colonies entirely. Researchers who built a dedicated imaging apparatus for photographing plates designed it specifically to eliminate glare and surface reflections while providing uniform, diffuse light from all sides.8Europe PMC. Inexpensive Apparatus for High-Quality Imaging of Microbial Growth on Agar Plates You do not need a custom imaging box for routine counting, but the principle matters: diffuse, even lighting from multiple angles reveals colonies more reliably than a single bright overhead source. A dark background behind the plate also helps, which is one reason Quebec colony counters use a dark field with gridlines.
If you are photographing plates for documentation or for later software-based counting, consistency in your imaging setup is critical. Variations in lighting angle, distance, and background color between photos will confuse both human reviewers and automated algorithms.
Automated Colony Counters and Software
Automated counters promise speed, and they deliver on that. But accuracy has been the persistent sticking point, especially without human oversight. One evaluation of an automated counting system across multiple bacterial species and concentrations found that uncorrected automated counts differed from manual counts by an average of about 60%, with only a moderate correlation to the manual results. After a technician visually reviewed and corrected the automated output, the difference dropped to under 2% and the correlation became very strong.2Europe PMC. Evaluation of an Automated System for the Counting of Microbial Colonies The catch is that the corrected automated approach actually took longer per plate (about 104 seconds) than pure manual counting (about 70 seconds), though the cognitive load on the technician was lower.
Machine-learning approaches are improving quickly. A deep-learning model based on an improved version of the YOLO object-detection algorithm achieved over 97% accuracy in detecting bacterial colonies, including small dotted colonies that traditional image processing tends to miss.9PubMed Central. A New Few-Shot Learning Method of Bacterial Colony Counting Based on the Edge Computing Device Another recent study comparing a YOLOv8 model to a conventional ImageJ-based workflow found that the deep-learning system detected about 35% more colonies than the manual image analysis process, largely because it could pick up subtle, low-contrast colonies and handle plate artifacts more robustly.10bioRxiv. AI-Based Detection of Coliform Colonies Using CNN Transfer Learning for Application to Cultured Plate Analysis in Water Quality Research
Smartphone apps for colony counting also exist, though performance varies. A study comparing four such apps found that only one showed high accuracy, and that was primarily at lower colony counts.11ScienceDirect / Journal of Microbiological Methods. Performance of four bacterial cell counting apps for smartphones For casual or field use they might be acceptable, but for anything regulatory or clinical you would want a validated system.
What Colony Counts Inherently Miss
A colony count gives you the number of colony-forming units, not the total number of living bacteria. Those are different numbers, sometimes by a wide margin. Several factors contribute to the gap.
First, clumping. If two or more cells stick together and are not separated during sample preparation, they will grow into a single colony. Your count records that as one CFU even though it originated from multiple cells. This is why the correct term is “colony-forming units” rather than “cells” or “bacteria.”
Second, and more significantly, many bacteria can enter a viable but non-culturable (VBNC) state. These cells are alive and metabolically active, but they will not grow into colonies on routine agar. This means conventional plate counting underestimates the total viable population in a sample, which is a recognized concern in both environmental monitoring and clinical microbiology.12PubMed Central. The importance of the viable but non-culturable state in human bacterial pathogens Flow cytometry and molecular methods can detect these cells, but those are complementary techniques, not replacements for plating.13PubMed Central. Specific and rapid enumeration of viable but nonculturable and viable-culturable gram-negative bacteria by using flow cytometry
Third, your medium and incubation conditions select for organisms that thrive under those specific conditions. Bacteria that need different nutrients, a different temperature, or an anaerobic environment will not grow and will not be counted. No single set of plating conditions captures everything in a mixed sample.
Using Selective and Chromogenic Media
When you only care about one type of organism in a mixed sample, selective or chromogenic media can make counting far easier. These media contain ingredients that suppress the growth of unwanted organisms while allowing the target to grow, and chromogenic versions include enzyme substrates that cause target colonies to turn a distinctive color.
For example, chromogenic plating media for pathogenic Listeria species use substrates that are cleaved by a specific virulence-related enzyme, causing Listeria monocytogenes colonies to appear as a different color from background flora. These media allow detection and enumeration of pathogenic Listeria within 24 to 48 hours, without the multi-step confirmation testing that traditional methods require.14PubMed. New chromogenic plating media for detection and enumeration of pathogenic Listeria spp.–an overview Similar chromogenic options exist for E. coli, Salmonella, MRSA, and other clinically or industrially important organisms.
When using selective media, keep in mind that your count represents only the target organism, not the total bacterial load. If you need both the total count and the count of a specific organism, you will need at least two different plates from the same dilution series.
Statistical Confidence and How Much Your Count Can Vary
A colony count is a sample-based estimate, and like any estimate, it has uncertainty. The distribution of colonies across plates from the same dilution follows a Poisson distribution for most food and environmental samples, meaning the variance equals the mean.15PubMed. The Validity of the Poisson Distribution to Analyze Microbial Colony Counts on Agar Plates for Food Samples In practical terms, this means that a plate with 100 colonies has an expected standard deviation of 10, giving you roughly a ±20% confidence interval at the 95% level. A plate with only 25 colonies has proportionally wider uncertainty. This is the statistical reason for preferring plates in the higher portion of the countable range when possible.
Replicates help. Plating the same dilution in duplicate or triplicate and averaging the counts tightens your estimate. Most regulatory methods call for at least duplicate plating at each dilution. If your duplicate plates differ by more than about 15 to 20% from their mean, that usually signals a technique problem (uneven spreading, pipetting errors, or inadequate mixing of the dilution before plating) rather than normal statistical variation.
Alternatives to Plate Counting
Colony counting is the workhorse of microbiology, but it is far from the only way to quantify bacteria. Other methods can complement or, in some cases, replace plate counting depending on the application.
Impedance microbiology measures changes in the electrical properties of a growth medium as bacteria multiply. It is used in dairy testing, among other applications, for rapid estimation of bacterial content in milk.16PubMed. Impedance microbiology: quantification of bacterial content in milk by means of capacitance growth curves The technique does not produce a colony count, but it can estimate microbial load much faster than waiting for colonies to grow. Similarly, turbidity measurements (optical density), quantitative PCR, and flow cytometry each offer speed advantages or the ability to detect non-culturable cells, though each comes with its own limitations and calibration requirements.
For most routine work in food safety, water testing, and clinical labs, the plate count remains the gold standard. It is simple, inexpensive, and directly shows you that the bacteria are alive and able to grow. The key to good results is careful dilution, consistent plating technique, appropriate lighting for counting, and an honest assessment of which plates fall in a range where your count is trustworthy.