Microbial Limit Determination: Counts and Three-Class Sampling Plans
How Microbial Limits Are Judged
Microbiological results (total plate count, coliforms, pathogens and similar) are compared with the limits in the product specification to decide whether a lot is acceptable. Results are usually reported as CFU/g (mL) following standards such as GB 4789, and limits are often expressed as a three-class sampling plan (n, c, m, M): n is the number of samples tested, c is the allowable number of samples exceeding m, m is the acceptable limit, and M is the unacceptable limit. The three-class plan reflects the spread of microbial counts within a lot better than a single-value comparison.
How to Apply a Three-Class Plan
Test n samples individually: the lot is acceptable when all results are <= m, or when the number of results between m and M is <= c. The lot is rejected when any single result exceeds M, or when more than c samples fall between m and M. Enter the n sample results together with m, M and c and the tool judges automatically, supporting total plate count, coliforms, molds and yeasts. Make sure all results use the same units of CFU/g (mL).
How to Handle Colony Counts
For the plate count method, select a suitable dilution (30 to 300 CFU per plate) to calculate the total viable count. The tool converts counts from multiple dilutions to CFU/g automatically and handles special cases such as not detected (less than the reciprocal of the lowest dilution) and spreading colonies, reporting results with the appropriate significant figures. Coliform MPN results can be combined with the MPN tool. Always count from a suitable dilution so that unreliable counts do not lead to a wrong judgment.
Precautions
Confirm the limit standard for the product category before judging, since limits differ by matrix (for example, ready-to-eat versus raw products). Re-test when results are close to m or M. Pathogen limits are usually not detected (n = 5, c = 0, m = 0) and are judged as a two-class plan. Because microbial results vary widely, re-check results near the limit, pay attention to sampling representativeness, and retain all original records.