9: Picking the plate to count

thumbs upFor this, we have any easy rule of thumb: the plate you count should have between 30 and 300 colonies. Use your mouse below to see how this works.

This rule of thumb has the undeniable advantage of being easy to use. But why does it work?

Let’s imagine that we have a sample with exactly 42 million cells per mL. And let’s say that our technique is flawless. If we count a plate prepared with 1 mL of a 1:1,000,000 dilution, we should find exactly 42 CFUs (remember, we have magically flawless technique and the medium and growth conditions suit the bacteria in the sample). But we would get the same number of colonies if our sample started with 42,000,001 cells, or 42,000,100 cells. Or even 42, 100,000 cells. In fact, even with our absolutely flawless technique, the best we can say is that there are between 41,500,000 and 42,500,000 cells per mL in the original sample. So our absolutely unavoidable error is plus or minus 500K/42 million — about 1.1%. That’s pretty good.

Let’s look at some other ways we could count the same 42 million cell sample — pay special attention to the “Error” column.

Dilution Factor plate count error ease of counting

(or 10-4)


41,005,000 to 41,005,000

+/- 5000 = 0.011%


NO, can’t do it, this is too much to count

(or 10-5)


41,050,000 to 41,050,000

+/- 50,000 = 0.11%

still great

Takes time, but countable

(or 10-6)


41,500,000 to 41,500,000

+/- 500,000 = 1.1%

not too bad

Easy to count

(or 10-7)


35,000,000 to 45,000,000

+/- 5 million= 11%

ACK! mayday!

A four year old can count this plate!


The information in the table shows that the theoretical error would be less if we used samples from lower dilutions — but the colonies would be so crowded, it would be impossible to count them accurately. Overcrowding also means that not all viable cells in the sample will form colonies because of the competition for space and nutrients. One the other hand, if we base our calculations on plates with very few colonies, the error is too great

The more we dilute, the easier the counting, but the higher the error. So you can see that the 30 to 300 guideline is really a compromise between countability and accuracy.