In the absence of a three-dimensional structure for TCR molecules, several attempts to identify their hypervariable regions by variability methods have been made; this subjects is still troublesome. In this paper three different variability indexes were used: (i) the Kabat index, which is the classical measure of sequence variability, (ii) the modified Kabat index, successfully used in the beta-chain of T-cell receptors and (iii) an information-theoretical entropy concept, recently proposed as an improved measure of the variability. In order to identify the hypervariable regions in the TCR sequences, a Fourier filtering was applied on each variability profile. Results show that the three variability indexes have distinct resolutions for different levels of variability. Thus, the simultaneous use of these indexes compensates for the deficiency of any one of them in estimating variability. Applying the Fourier filtering, it is found that the hypervariable regions here identified, roughly coincide with the defined CDR-2 and CDR-3 in TCR by analogy with Ig. However, no hypervariable in the CDR-1 of alpha- and beta-chains was found. The study on the influence of sample size in variability analysis, indicates that results are independent of the sample size. Considering current structural models of TCR-peptide-MHC interaction, one can suggest that the low-variability characteristics of these regions is inherently related to the interaction with relatively conserved region on the alpha-helices of MHC.