The atmosphere in which the test system is immersed can have a major effect. Anaerobic organisms cannot grow in the presence of oxygen, and tissue cultures may require the presence of 5% CO₂ to grow well. Certain facultative organisms will adjust their metabolic paths to cope with reduced levels of oxygen. This, in turn, can affect their growth rates. When media for general purposes, such as sterility tests, are being considered, it is normal to include one medium that provides anaerobic conditions. The detection of anaerobes is important as they include toxin-producing and other pathogenic bacteria.

Quantitative Issues One of the problems with quantitative microbiological tests is that as microbe counts become smaller, straight-forward linear behavior is less common than that which follows the Poisson distribution. This is because random distribution is not even distribution. Most quantitative tests for microorganisms require the plating of dilute liquid samples, and it is normal to prepare samples to ensure the dispersion of microbes and a random distribution of bacteria or viruses. When concentrations are high, the lack of even distribution is not a problem; simple linear averaging methods can compensate for the uneven distribution. Problems arise with smaller numbers of microbes.

Consider an example where there are exactly 100,000 organisms per mL. If 0.1 mL is taken and mixed with 0.9 mL of a diluent, it is highly unlikely that the new suspension will contain exactly 10,000 organisms; it would not be surprising to have anywhere from 9,800 – 10,200 organisms. Back-calculating the result produces a range from 98,000 – 102,000 organisms in the original sample, and, if there were enough replicates, the results could be averaged to obtain a number indistinguishable from 100,000. This is the result that would be expected based on linear thinking.

A transition occurred from a high density that produces a fairly smooth, homogeneous distribution of organisms to a low density that results in organisms that are distributed with significant distances between them. Under these conditions, the suspension behaves according to the Poisson distribution and assumptions related to a normal distribution no longer hold. The Poisson distribution is an exponential function. The problem is that parameters such as the standard deviations may be logarithmic in nature, and when attempts are made to make these numbers "real" by taking the antilogarithms, the results may actually have no "real" meaning. This can cause great difficulties when attempting to validate quantitative microbial test procedures.

When it is necessary to deal with the Poisson distribution, it is wise to consult a statistician who is versed in the use of this distribution. It appears that the transition to the Poisson distribution occurs when approximately 100 colonies or plaques are counted. This is unfortunate because at this level many analysts will declare a colony or plaque count to be "too numerous to count" (TNTC) to avoid the tedium of these measurements. Therefore, most colony or plaque counting procedures actually operate under the Poisson distribution and calculations based on the normal distribution will be incorrect.

Revalidation The frequency of revalidation is a contentious question. There are many tests, such as the growth promotion test on culture media, that are essentially self-validating and are run frequently. It could be argued that if performance parameters (for example, percent recovery of indicator organisms) are monitored via control charting and no significant changes are seen, revalidation is unnecessary. However, control charting usually does not measure all the parameters included in validation studies. Consequently, it is wise to revalidate tests after any major change in constituents or procedures; in fact, revalidation may be needed to justify the changes. Changes in suppliers (especially of media components) and changes in the composition of test samples have resulted in major changes in microbiological tests. Finally, it is probably wise to revalidate procedures approximately every second year to protect against unseen or unreported changes. A media supplier may change its own suppliers or change its processing procedures without notifying customers. The supplier may have no idea of the impact these changes could have on the end use of their product. In addition, personnel changes in the laboratory and the maturing of analysts' techniques can have an effect.