Counting bacterial colonies on an agar plate comes down to visually inspecting each distinct colony, tallying the total, and then working backward through your dilution factor to estimate how many bacteria were in the original sample. The count itself is deceptively simple: you look at the plate, you count every visible colony, and you multiply. But getting an accurate, reproducible number depends on choosing the right plates to count, recognizing what is and is not a real colony, and understanding the limits of what a colony count actually tells you.
The Actual Counting Process
Pick up your incubated plate, flip it over so you are looking through the bottom of the dish, and count every distinct colony you can see. Most people use a felt-tip marker to dot the bottom of the plate as they count each colony, which prevents double-counting. A colony counter with a magnifying lens and grid background helps when colonies are small or faint, but a pen and your eyes work fine for plates that are not too crowded. Count every colony regardless of size or morphology unless you are specifically looking for one organism on a selective medium.
Work systematically. Pick a section of the plate, count left to right or top to bottom, and mark as you go. If colonies are packed close together, slow down. Two colonies sitting side by side with a visible boundary between them count as two. A single mass of growth with no visible separation counts as one. When two or more colonies overlap but you can still see where one ends and the other begins, count each one separately.
Which Plates Are Worth Counting
Not every plate in your dilution series gives you a usable number. The widely taught guideline is that a plate should have between 30 and 300 colonies to be considered countable. Below 30, random variation has too much influence on your result. Above 300, colonies start merging and crowding each other, which leads to underestimation because you cannot distinguish individual colonies anymore. Some protocols use a tighter range, and some regulatory methods shift the window depending on the organism, but the 30-to-300 range is the standard starting point for most laboratory work.
If you plated multiple dilutions, pick the plate (or plates) that fall within that countable range. If more than one dilution gives you a plate in the sweet spot, count both and average the results after adjusting for each dilution factor. If none of your plates land in the countable range, you have a problem: either your original sample was more concentrated or more dilute than you expected, and you may need to repeat the experiment with adjusted dilutions.
Calculating Colony-Forming Units
The number you get from counting colonies is not the final answer. What you are really after is the concentration of viable bacteria in your original sample, expressed as colony-forming units per milliliter (CFU/mL). The formula is straightforward: take the colony count, divide it by the volume you plated (in milliliters), and divide again by the dilution factor for that plate. If you counted 150 colonies on a plate where you spread 0.1 mL of a 1-in-10,000 dilution, the concentration in your original sample is 150 divided by 0.1 divided by 0.0001, which gives 15,000,000 CFU/mL.
Serial dilution methods are the standard way to get bacterial cultures into the countable range, and from those colony counts, researchers infer bacterial concentrations measured in CFU.1Europe PMC. Maximum likelihood estimators for colony-forming units The term “colony-forming unit” rather than “cell” is deliberate. A single colony might have grown from one bacterium, or from a clump of bacteria that were stuck together when they landed on the agar. The count reflects the number of viable units that could produce visible growth, not the exact number of individual cells in the sample.
Pour Plates vs. Spread Plates
The two most common quantitative plating methods are the spread plate and the pour plate, and they look different when it comes time to count. In a spread plate, you pipette your diluted sample onto the surface of solidified agar and spread it evenly with a glass or plastic spreader. Colonies grow on top of the agar surface and are easy to see. In a pour plate, you mix your sample with molten agar (cooled to about 45-50°C so it does not kill the bacteria), pour the mixture into the dish, and let it solidify. Colonies then grow throughout the agar, both on the surface and embedded within it.2Europe PMC. Aseptic laboratory techniques: plating methods
Counting pour plates takes more attention because subsurface colonies look different from surface colonies. They tend to be smaller, lenticular (lens-shaped), and sometimes harder to distinguish from air bubbles or debris trapped in the agar. One comparative study found differences in colony size and morphology between pour plates and spread plates for the same organism.3PubMed. Use of the pour plate technique in tuberculocidal efficacy testing according to EN 14348 – a comparative study When you are counting a pour plate, rotate the plate and look at it from different angles. Subsurface colonies appear as small dots or discs floating within the agar, and they should not be confused with trapped air bubbles, which tend to be perfectly round and do not grow over time.
Problem Plates and How to Handle Them
Some plates look straightforward but are not. Here are the most common complications and what to do about them.
Swarming organisms like Proteus species spread across the plate in waves of motile growth, covering other colonies and making individual counts impossible. On a standard nutrient agar plate, a single Proteus colony can expand to cover a large area within hours. Various methods exist to suppress this swarming, including adding specific chemicals to the medium or adjusting the agar concentration.4Europe PMC / American Society for Microbiology. Abolition of swarming of Proteus If you are working with clinical samples where Proteus is a possibility, use a medium formulated to inhibit swarming, or your colony counts will be unreliable.
Spreaders, which are colonies that have spread out due to excess surface moisture, cause similar headaches. If moisture collects on the agar surface before or during plating, some organisms will glide across the wet surface and produce a film rather than a discrete colony. Drying your plates in an incubator with the lids slightly ajar before use reduces this problem.
Artifacts like air bubbles, scratches in the agar, and particulate contamination can fool both humans and machines. A research team developing deep-learning tools for colony detection noted that surface impurities from spreading the sample, evaporating air bubbles in the agar, and light speckles were all commonly misidentified as colonies and had to be filtered out using specialized algorithms.5Light: Science & Applications. Early detection and classification of live bacteria using time-lapse coherent imaging and deep learning For manual counting, the practical test is whether the dot in question looks like it has grown over time and whether it has the expected morphology for a bacterial colony: slightly raised, opaque, and with defined edges.
Counting on Chromogenic and Selective Media
When your agar contains chromogenic substrates or selective agents, counting becomes both easier and harder. Chromogenic media produce colonies of specific colors depending on the enzyme activity of the organism. This means you can count not just total colonies but colonies of a particular type based on color alone. Automated systems can take advantage of this by applying color filters to digital images of the plates. One such device, the ScanStation, tracks real-time colony growth on chromogenic agar and lets users sort plates by the presence or absence of colonies of specific colors.6PubMed Central. Rapid and automated screening of carbapenemase- and ESBL-producing Gram-negative bacteria from rectal swabs using chromogenic agar media and the ScanStation device
The tricky part is that the color intensity of colonies can vary with incubation time, colony size, and the density of plating. A pale green colony at 18 hours might be a vivid green colony at 24 hours, so reading chromogenic plates too early can lead to misclassification. If you are counting colored colonies manually, follow the manufacturer’s incubation time closely and compare colony colors against reference images when available.
How Much Human Error Creeps In
Manual colony counting is subjective. Two trained microbiologists counting the same plate will not always get the same number, and the discrepancy grows as the plate gets more crowded or the colonies get smaller. An evaluation of an automated counting system found that manual counts served as the reference standard, but even human counters introduced variation that had to be accounted for when benchmarking the machine.7Europe PMC / ASM Journals (Microbiology Spectrum). Evaluation of an Automated System for the Counting of Microbial Colonies In that study, automated counting without any human correction differed from manual counts by about 60% on average, which is a huge gap. But when a human reviewed and corrected the automated results, the difference dropped to under 2%, and the correlation with manual counts was nearly perfect.
The time tradeoff is interesting. Manual counting averaged about 70 seconds per plate. Fully automated counting with no correction took about 30 seconds. Automated counting followed by visual correction averaged 104 seconds, actually longer than just counting by hand.7Europe PMC / ASM Journals (Microbiology Spectrum). Evaluation of an Automated System for the Counting of Microbial Colonies For high-volume labs processing hundreds of plates a day, the slight per-plate slowdown may still be worth it for consistency and record-keeping. For a research lab counting a dozen plates, manual counting with a marker and a steady hand remains perfectly adequate.
Automated Colony Counters and Software
Desktop software and dedicated instruments for automated colony counting have improved substantially in recent years. These systems photograph the plate, apply image-processing algorithms to identify individual colonies, and report a count. Some use traditional segmentation approaches, where the software looks for circular objects that are brighter or darker than the background. Others use machine-learning models trained on thousands of annotated colony images.
One automated tool called MCount achieved an average error rate of about 4% when tested against ground-truth counts on sub-images of agar plates, substantially outperforming several other algorithms that had error rates ranging from roughly 17% to over 50%.8PubMed Central. MCount: An automated colony counting tool for high-throughput microbiology The biggest weakness shared by many automated systems is handling crowded plates where colonies touch or overlap. One segmentation tool, AutoCellSeg, uses a watershed algorithm that splits touching objects along intensity valleys, but it tends to count a continuous region of merged colonies as a single colony, which leads to severe underestimation on dense plates.9Scientific Reports. AutoCellSeg: robust automatic colony forming unit (CFU)/cell analysis using adaptive image segmentation and easy-to-use post-editing techniques
Deep-learning approaches have started to address some of these limitations. One team trained a modified object-detection model to recognize bacterial colonies of different sizes and shapes, including tiny dot-like colonies and larger circular ones, achieving accuracy above 97% with a false-negative rate of just 2%.10PubMed Central. A New Few-Shot Learning Method of Bacterial Colony Counting Based on the Edge Computing Device Training these models requires large annotated datasets. One publicly available dataset includes images from 24 different bacterial species cultured at various concentrations on solid media, providing the kind of diversity needed for robust model training.11PubMed Central. Annotated dataset for deep-learning-based bacterial colony detection
Smartphone Apps for Colony Counting
If you do not have access to a dedicated colony counter, several smartphone apps promise to count colonies from a photo. A head-to-head comparison of four such apps found wide variation in performance. The best-performing app, CFU.Ai, correlated well with manual counts on blood agar and LB agar, with accuracy metrics above 0.97 on both media types. The second-best app, Promega Colony Counter, performed respectably on blood agar and chromogenic agar but struggled more on LB. The other two apps tested showed poor accuracy across the board, with one essentially failing to detect colonies at all on certain media.12Journal of Microbiological Methods. Performance of four bacterial cell counting apps for smartphones
A common pattern was that sensitivity dropped as colony counts got higher, meaning the apps missed more colonies on crowded plates, exactly the situation where you most need help counting. Some apps also falsely identified features on the plate rim or artifacts as colonies, inflating counts at low concentrations. The researchers concluded that none of the tested apps could fully replace manual counting, though the better ones could serve as a useful reference or provide semi-quantitative estimates.12Journal of Microbiological Methods. Performance of four bacterial cell counting apps for smartphones If you use an app, treat it as a starting point and eyeball the result against what you see on the plate. If the app’s number seems off by more than a handful, count manually.
What Colony Counts Cannot Tell You
Every colony count rests on a fundamental assumption: that the bacteria in your sample can grow on the medium you used, under the conditions you incubated. Bacteria that are alive but unable to form colonies under your specific conditions will be invisible. Some cells enter a stressed state after exposure to heat, acid, antibiotics, or starvation. These cells are metabolically active but will not divide on standard agar. This means colony counts can significantly underestimate the true number of live bacteria in a sample.13Europe PMC. Revisiting Strategies for Bacterial Enumeration: The Case for Viable but Nonculturable (VBNC) Cells in Supplement Products
This matters especially for environmental samples, food safety testing, and probiotic products, where the organisms of interest may have been through processing conditions that push cells into that stressed state. If your application demands a total live-cell count rather than a culturable-cell count, you may need to supplement plate counts with methods like fluorescence microscopy or flow cytometry, which detect cells based on membrane integrity or metabolic activity rather than growth.
Counting Yeasts and Molds
Colony counting is not limited to bacteria, and the principles transfer to fungi with some adjustments. Yeasts typically form discrete, often creamy or pasty colonies that are easy to count. Molds are trickier because they produce filamentous growth that can spread across the plate. A single mold colony can cover a large area with aerial hyphae, making it difficult to tell where one colony ends and the next begins. Specialized media like Petrifilm yeast and mold plates use indicator dyes and nutrient-binding systems that keep mold colonies compact, making them far easier to count. Validation studies have shown that the vast majority of yeast and mold species produce typical countable morphology on such plates, while most non-target bacteria are suppressed.14Oxford University Press. 3M™ Petrifilm Yeast and Mold Count Plate for the Enumeration of Yeasts and Molds in Dried Cannabis Flower: AOAC Official Method SM 997.02
When counting mold colonies on standard agar, work from the center of each colony outward and look for distinct points of origin. Two mold colonies whose edges have grown together will often have two separate dense centers with a boundary zone between them. Count each center as one colony. If a mold has sporulated heavily, the colored spore mass can help you see the center of each colony more clearly.
Practical Tips That Save Repeat Experiments
A few small habits make a big difference in getting counts you can trust. Plate at least two dilutions, and ideally three, so that at least one lands in the countable range even if your estimate of the sample concentration is off. Always plate in duplicate or triplicate. Two plates from the same dilution that give wildly different counts tell you something went wrong with your technique, whether the sample was not mixed well or the spreader did not distribute cells evenly.
Label the bottom of every plate, not the lid. Lids get swapped. Include the dilution, the date, and the sample identity. Count your plates at a consistent time after inoculation, because colony size increases with incubation time and late counts on crowded plates will show more merging than early counts. If you cannot count a plate immediately, refrigerate it to slow further growth, but do not leave it for days.
Use a dark background for light-colored colonies and a light background for pigmented ones. Oblique lighting, where you angle the plate so light hits it from the side, makes tiny colonies pop out against the agar surface. And if a plate has a colony count anywhere near 300, take extra time. Those borderline-crowded plates are the ones where people most often lose track and either double-count or skip colonies.