From more than one hundred years to now, natural rubber (NR) has been described as an enigmatic, rare and expensive material. Automobile industry, which owes a lot to NR (and vice versa), has promoted its production (7 millions metric tons in 2002). This agro-material with specific properties, still unequalled until now for technical items manufacturing, presents a handicap for users: its lack of consistency. Indeed, with automation of manufacturing processing, industrials seek raw natural rubber with constant quality from one lot to another such as it is for synthetic rubbers. It is thus crucial to be able to produce constant quality natural raw rubber - which is not the case today - by identifying new physicochemical criteria more pertinent than the currently used ones. Actually, national (STR, SMR, SIR, ...) or international (ISO 2000) standards which were created for this purpose are useful but not sufficient to characterise precisely enough the technological properties of raw NR. This concern, more often raised by north manufacturers, may also become a reality for manufacturers from NR-producing countries. Moreover, the synthetic rival of NR could be strengthened if the announced 2005 shortage of NR occurs in the elastomer market. Thus a main challenge for natural rubber technologists remains to improve the prediction of NR manufacturing behaviour. In this paper, a strategy, based on the relationship between structure of NR and manufacturing behaviour, is proposed in order to find new criteria more relevant to predict NR behaviour. It has to be performed from initial stages (from tapped tree to manufacturer's mixer) to downstream (product behaviour and variation during the manufacturing (blending, shaping, vulcanisation)). Indeed, to obtain applicable results; this approach should be as integrated as possible and thus should involve researchers (agronomists, technologists, biochemists, etc...) and industrial partners (raw rubber producers and end-used product m