Three methods are commonly employed to measure phenotypic mutation rates: mutation accumulation assays, mutant accumulation assays, and fluctuation assays.

The mutation accumulation assay involves passing a culture through recurrent bottlenecks, ideally of a single cell/individual, such that all mutations are nearly neutral. This is useful for determining the rate of mutations effecting fitness since repeated bottlenecks will reduce the effect of selection (KIBOTA and LYNCH 1996; ZEYL and DEVISSER 2001). This method works well in multicellular organisms, where the population size can be maintained at the bottleneck; however, in microorganisms, where a visible colony must be allowed to form, selection will still occur between the bottlenecks. Several methods are available for estimating phenotypic mutation rates from mutation accumulation assays (GARCIA-DORADO and GALLEGO 2003); alternatively, direct sequencing can be used since all mutations occur in the same genome (DENVER et al. 2004).

In the mutant accumulation assay the frequency of a neutral phenotype is monitored in an exponentially growing culture by periodically plating an aliquot of the culture onto selective media. Once the population reaches a size such that the probability of a new mutation occurring in the next generation is approximately one, the frequency of mutants will increase linearly with time. An accurate estimate of phenotypic mutation rate requires a long period of time between frequency measurements, making these experiments vulnerable to beneficial mutations, which are more likely to occur in the non-mutant population and slow the accumulation of mutants.

In the fluctuation assay many parallel cultures are inoculated with a small number of cells, grown to saturation under non-selective conditions, and plated to select for mutants (LURIA and DELBRUCK 1943). The number of mutations that arise in each culture will follow the Poisson distribution; however, the number of mutant cells per culture will vary greatly since early mutations will lead to "jackpots," cultures that contain a great many mutant individuals. The simplest way to estimate the expected number of mutations that occur in each culture (m) is from the fraction of cultures with zero mutants, which should be equal to e^-m. This method (P0) was used by LURIA AND DELBRUCK (1943) in the original paper describing the fluctuation assay. The full distribution of mutants per culture (the Luria-Delbruck distribution) can be described by a set of recursive equations (MA et al. 1992). The most accurate method for estimating m (Ma-Sandri-Sarkar maximum likelihood) finds the m that gives the best fit of the Luria-Delbruck distribution to the data (ROSCHE and FOSTER 2000; SARKAR et al. 1992).