Systems so-called 'electronic noses' encountered many problems when they ca
me out from research laboratories to plants. After improving many steps in
the systems, one of the remaining problems is the limited range of the sign
al-to-noise ratio which does not enable to maintain an efficient search in
the library patterns of the products database. This can be solved by the cl
assical way, which consists in increasing the sensitivity of the sensing el
ement itself, or by an alternative way which is increasing the amount of vo
latile compounds reaching the sensitive element. After providing some backg
round information about the measurement of food aroma, we will review the m
ain methods that can be found in electronic olfactometers for managing the
effluent transfer. In this study, we have only compared the usable transfer
methods, independently of the kind of sensors used. Five transfer methods
were compared versus the fully static original method. Advantages and drawb
acks were studied for each method. The main parameters were: efficiency, im
portance of transients (kinetic information), automation and sensor recover
y time (level of pollution). Obviously, no method was perfect, but each hav
e a lot of pros and cons. This work indicates that the transfer technique c
onsiderably influences the system performance. We have also proposed a nove
l approach based on the headspace kinetic information for improving the cla
ssification of samples which are relatively close in terms of flavour-or wh
ich contains major compounds of weak flavour impact but present at relative
ly high concentrations, This method. still under development, seems to be p
romising in terms of efficiency (sensitivity) and transient analysis. (C) 1
999 Elsevier Science S.A. All rights reserved.