In a previous study the sounds emitted by a mother cat and those emitted by the kittens during the first weeks of life were compared. It found that the fundamental frequency of the kitten vocalisation is two or three times the value of the fundamental frequency of their mother. The aim of this study is to describe the harmonic structure of the sound, the variation of values of the fundamental frequency during the somatic growth and to study the experimental pattern of Flanagan (1972). The formant frequencies depend, in a complex way, on the dimensions and the shape of the vocal tract. The contact vocalisations (mews) were elicited by removing kittens from their mothers. The vocalisations of four different families of cats were collected and analysed from the second day until the 7th month of the kitten's lives (total number of kittens = 13). Recordings were made with an analogue tape recorder Marantz CP 230) and two digital ones (Aiwa HD-SI and Casio DA-1); connected microphones were Sennheiser ME88. Vocalisations were analysed with the program Voxys installed in an IBM compatible computer equipped with a 16-bit Sound Blaster card. The fundamental frequency of each vocalisation was computed and the T0 period obtained by cepstrum analysis at the point of maximum stability of the signal. This was obtained by searching the zone of temporal stability during the emission of the sound in three subsequent frequency bands. The formants were obtained by cepstral analysis of the spectrum at the point of maximum stability. Comparing kittens' mews with those of the mother, it is evident that during the early weeks of life, the fundamental frequency (1?0) of the kittens' vocalisations is from 2 to 3 times the fundamental frequency of their mother's. For each sound of the kitten the value of F0 decreases as a mean trend with age, while it is constant in each sound of an adult cat. When the kittens are 60 days old the F0 is comparable to their mothers' F0 (the maximum difference recorded is 80 Hz). The ratio among the first three formants is 1:2:3. According to Flanagan's model this ratio indicates that the shape of the vocal tract is similar to a cylinder with the aboral extremity open. The measurements of F0, F2 and F3 enabled the discrimination of the mother's vocalisations with an accuracy of 100% correctly classified cases (discriminant analysis: Wilks lamda; F = .17. 155, P = 0.000). The vocalisations of 15 days o1d kittens are correctly classified to their families in percentages varying from 78 to 100%. Similar results are obtained with 30 days old kittens. At 60 days the percentage of correctly classified vocalisations reaches 99%.