In biology, phenetics (/fɪˈnɛtɪks/; from Ancient Greek φαίνειν (phainein) 'to appear'), also known as taximetrics, is an attempt to classify organisms based on overall similarity, usually with respect to morphology or other observable traits, regardless of their phylogeny or evolutionary relation. It is related closely to numerical taxonomy which is concerned with the use of numerical methods for taxonomic classification. Many people contributed to the development of phenetics, but the most influential were Peter Sneath and Robert R. Sokal. Their books are still primary references for this sub-discipline, although now out of print.

Phenetics has been largely superseded by cladistics for research into evolutionary relationships among species. However, certain phenetic methods, such as neighbor-joining, are used for phylogenetics, as a reasonable approximation of phylogeny when more advanced methods (such as Bayesian inference) are too expensive computationally.

Phenetic techniques include various forms of clustering and ordination. These are sophisticated methods of reducing the variation displayed by organisms to a manageable degree. In practice this means measuring dozens of variables, and then presenting them as two- or three-dimensional graphs. Much of the technical challenge of phenetics concerns balancing the loss of information due to such a reduction against the ease of interpreting the resulting graphs.

The method can be traced back to 1763 and Michel Adanson (in his Familles des plantes) because of two shared basic principles – overall similarity and equal weighting – and modern pheneticists are sometimes termed neo-Adansonians.

Phenetic analyses are "unrooted", that is, they do not distinguish between plesiomorphies, traits that are inherited from an ancestor, and apomorphies, traits that evolved anew in one or several lineages. A common problem with phenetic analysis is that basal evolutionary grades, which retain many plesiomorphies compared to more advanced lineages, seem to be monophyletic. Phenetic analyses are also liable to be rendered inaccurate by convergent evolution and adaptive radiation. Cladistic methods attempt to solve those problems.

Consider for example songbirds. These can be divided into two groups – Corvida, which retains ancient characteristics of phenotype and genotype, and Passerida, which has more modern traits. But only the latter are a group of closest relatives; the former are numerous independent and ancient lineages which are related about as distantly to each other as each single one of them is to the Passerida. For a phenetic analysis, the large degree of overall similarity found among the Corvida will make them seem to be monophyletic too, but their shared traits were present in the ancestors of all songbirds already. It is the loss of these ancestral traits rather than their presence that signifies which songbirds are more closely related to each other than to other songbirds. However, the requirement that taxa be monophyletic – rather than paraphyletic as for the case of the Corvida – is itself part of the cladistic method of taxonomy, not necessarily obeyed absolutely by other methods.

The two methods are not mutually exclusive. There is no reason why, e.g., species identified using phenetics cannot subsequently be subjected to cladistic analysis, to determine their evolutionary relationships. Phenetic methods can also be superior to cladistics when only the distinctness of related taxa is important, as the computational requirements are less.

The history of pheneticism and cladism as rival taxonomic systems is analysed in David Hull's 1988 book Science as a Process.