The aim of this paper is to analyse the relationship between self-report hedonic evaluations and the physiological expression of emotion in response to odorants. We try to solve the following questions: (1) Is it possible to find any experimental evidence that the sense of smell is linked with emotion? (2) What kind of odorants can be distinguished by autonomic analysis? (3) Is there a link between hedonics and autonomic information? The effects of odorants on the emotional process were estimated, in terms of autonomic nervous system (ANS) activity. Fifteen subjects inhaled five odorants as olfactory stimuli: lavender (LAV), ethyl acetoacetate (EAA), camphor (CAM), acetic acid (AA) and butyric acid (BA). After inhaling the odorant, subjects were requested to fill out an 11-point hedonic scale to rate its pleasantness versus unpleasantness. ANS parameters were as follows: two electrodermal responses, skin potential (SP) and resistance (SR); two thermovascular parameters, skin blood flow (SBF) and skin temperature (ST); and two cardiorespiratory parameters; instantaneous respiratory frequency (IRF) and instantaneous heart rate (IHR). Simultaneous recording of six parameters showed that specific autonomic patterns were associated with each odorant. An analysis of variance made it possible to differentiate among the five odorants. Two-by-two odorant comparisons for autonomic responses using Tukey's HSD multiple comparison test only permitted differentiation between pleasant odorants (LAV and EAA) and unpleasant (AA and BA) ones, but camphor was differentiated from both pleasant and unpleasant odorants. Each odorant elicited responses in the different parameters, yet subjects responded through their preferential channels; an average of two channels was used by each subject. These results when compared with those obtained with other senses (visual and auditory), did not evidence the postulated preferential link between olfaction and emotion. A strong link between hedonics and ANS response could be demonstrated when considering each subject and mainly through his/her preferential channel(s); conversely a weak correlation (SR duration excepted) was obtained between inter-subjects' hedonic evaluation. It seems that for a given population the autonomic response reflect the odor valence only through some parameters related to the main preferential channel(s) and thus the global autonomic pattern has to be considered.