Odor input evokes characteristic, time-evolving (non-stationary) events in the spontaneously active central ganglia of the snail Helix pomatia. Assuming stationarity for the signals, one could, as the first approach, apply the Fourier-based methods, frequency amplitude characteristics (FAC) measures, for analyzing such events. We could thus for the first time describe such events in frequency and amplitude and show that the frequency, at which power increases most, is specific to the odor or its class [Comp. Biochem. Physiol. 123A (1999a) 95; Comp. Biochem. Physiol. 124A (1999b) 297]. Wavelet tools assume no record stationarity and are suitable for describing the dynamically evolving brain electrical signals precisely and quantitatively. We, therefore, tested these tools for the typical odor experiments with the procerebrum (PC), the pedal ganglion (PG) and the visceral ganglion (VG) of the Helix, which we earlier analyzed by the FAC measures and compared both results. The two basic findings of the present wavelet analysis are as follows: (i) the wavelet energy fluctuations clearly visualize dynamical interactions among the major bands (0.1-3.1 Hz), implying a possible 'mutual exclusion' between slow components < 0.8 Hz and faster ones > 0.8 Hz. (ii) Entropy behavior was characteristically different for each of the three brain regions. Only in PC the response to aversive odorants (decrease of entropy = more ordered state) is differentiated in entropy from that to attractive ones (increase of entropy=more disordered or more complexly ordered state) indicating the odor-discriminating function of this region. In VG entropy of the intrinsic activity is so high (highly disordered state) due to the strong wideband activity reaching > 50 Hz that odor stimulation results mainly in lowering of entropy (= more ordered state) regardless of the nature of the odor. In PG, however, odor presentation generally increases entropy due to the robust, wide-band activation at > 3 Hz (sensorimotor function) that is generated as a secondary, but dominant and robust, response. In respect to describing time evolution of different frequency band components the present wavelet tools can much more sensitively do so, as compared with the FAC measures. They can also characterize a change in the neuroelectrical state in terms of entropy.