Abstract: Soluble proteins were extracted from mature leaves of 83 citrus biotypes. Their protein patterns were obtained by acrylamide gel electrophoresis. The similarity of the protein patterns among the samples was calculated by computer according to the principle and methods of numerical taxonomy. The similarity comparison was made between different species and varieties included in the Swingl’s system. The distance coefficients of similarity (DCS) is 2.13 between species in the genus Citrus, 0.927 in Poncirus and 0.617 among Fortunella species. The DCS of biotypes in Citrus limon, C. paradisi and C. sinensis are zero, and those in C. grandis and C. reticulata are 0.656 and 0.863 respectively. There is no difference among satsuma mandarin, Tankan and King mandarin, which are considered respectively as a good species in Tanaka’s system. Ancestors of some biotypes. whose origin was unknown berore, are postulated in this work. Jiang Bai Shang Cheng is probably a hybrid between C. junos Tanaka and mandarin, rough lemon may be a hybrid of mandarin with citron or lime, but not with lemon. Some so-called lemon biotypes are not in the same cluster as true lemon (Eureka or Lishon). The biotypes. of sour orange, all of which are scattered in the phenetic tree of cluster analysis (PTCA), might have been derived from different ancestors. In the PTCA the evolutionary tendency from Poncirus to Macroacrumen in the citrus plants may be found on the basis of the DCS of different genera, sections and subsections respectively with the subsection Macroacrumen. The present author considers that if Fortunella is recognized as true genus, Cephalocitrus should also be an acceptable genus. The position of Fortunella in the PTCA is between the sections Cephalocitrus and Aurantium, but it does not represent its position in the phylogeny. Taking comprehensive data into consideration it appear to be more reasonable to place C. ichangensis, as member of Papeda originated in the subtropics, in the section Papeda.

Key words: Citus, Protein, Electrophoresis, Similarity of the patterns, Cluster analysis.