Diversity studies have been made by numerous authors using several kind of descriptors: morphological, enzymatic, molecular. A clear classification is difficult to obtain due to several reasons: - the genetic mixing happened during the three last centuries, - the samples studied are sometimes biased: eg. the Forastero were collected by Pound in Peru for resistance to witches broom, - the first Criollo diversity studies were first based on genotypes present in germplasm and corresponding to hybrids rather than pure types, - the results about the several populations are widespread in several different studies. However the main results indicate: -Forastero populations. An important diversity is observed between and within UpperAmazon Forastero populations with a continue variation. The highest diversity has been observed in Ecuadorian populations LCTEEN and lower diversity in some populations as NA, GU (French Guiana populations). However, the studied samples from Upper Amazon are biased and prevent to compare really the extent of diversity between populations. The Ecuadorian Allen collections have been made in a larger area, without criteria of selection, and on a larger number of trees than the Peruvian Pound samples. Very few samples from Colombia and Brazil have been analysed in these studies. - Criollo and Nacional varieties. "Ancestral" Criollo and Nacional have been identified to be -nearly unique homozygous genotypes varieties. Modem Criollo and Nacional are in fact hybrid types resulting from introgression of respectively Forastero (reduced number of genotypes) in the ancestral Criollo and of Trinitario in the ancestral Nacional genotype. - The specificity and differentiation of some populations / types as the French Guiana Forastero population, the ancestral Nacional and Criollo varieties could be explained by foundation effects or refuge areas. The consequences for breeding programmes is that a reduced number of Upper-Amazon Forastero have been used until no