We have used the catalogue of (Taylor et al., 1993) to investigate the problem of a classification of pulsars. Five samples have been analysed by the main components method and by methods of the cluster analysis.

1. i) 15 pulsars with 14 known parameters P, Ṗ, W50, L (Malov et al., 1994), B, z,spectral index a, frequencies of maximum VM and minimum Vc in a spectrum (Malofeev et al., 1994), an angle between magnetic moment and rotation axis β (Malov, 1994), the transformation coefficient η = L/Ė, magnetic field near the light cylinder Blc and a maximum derivative of a position angle C = (dψ/dφ)max (Malov, 1990).

2. ii) 130 objects with 10 parameters (the spectral characteristics α, VM, VC and the parameter C were excluded from the consideration).

3. iii) 218 pulsars with 4 known parameters P, Ṗ, L and z. These parameters can be considered as independent variables.

4. iv) 89 sources with the same 4 parameters. This sample consists of two subsamples: 65 pulsars with P > 1 s and 24 ones with P < 0.1 s.

5. v) 24 pulsars with P < 0.1 s from the previous sample.