The proportion of molecular-based studies of population genetic struct
ure and taxonomic relationships is constantly increasing. The electrop
horesis of proteins is among the first molecular techniques that were
and continue to be widely used in systematics. Phylogenetic analysis o
f allozyme data is the real way to investigation of species and genera
relationships. However, direct utilization of allozyme data is hamper
ed by specific properties of loci such as-constant intrapopulation pol
ymorphism. Another problem concerns the determination of phylogenetic
significance of allele frequency information. Ambiguity of choice of c
haracters and rules of distinguishing between their states accounts fo
r variability of methods of phylogenetic inference used with allozyme
data. The current review presents an analysis of underlying assumption
s, virtues and drawbacks of the following methods: Hennigian cladistic
s, distance methods, maximum likelihood methods, frequency parsimony a
nd parsimony of discrete characters. For the latter method two models
treating as characters either loci or alleles are described, their rel
evance and principles of determination of character states and polarit
y are discussed. Properties of most popular genetic distances and crit
eria for their application with different distance methods are examine
d. The methods for testing the reliability of inferred trees, basic we
ighting procedures and perspectives for development of new algorithms
are considered.