Among the existing classifications in the field of psychiatry, Leonhar
d's comprises 35 forms of endogenous psychoses, allowing a more precis
e differentiation. Computer-assisted mathematical methods of numerical
taxonomy, frequently used in other fields of biological sciences, wer
e used to check its coherence, definitely proving its taxonomic validi
ty. Numerical values were assigned to signs and symptoms described by
Leonhard, taking into account their frequency and diagnostical importa
nce. The resulting data matrix was processed, obtaining the dissimilit
ude coefficients and the main components which validate Leonhard class
ification. This method can be applied to other psychiatric classificat
ions validating them or displaying their weaknesses.