Analysis of seismic ambient vibrations is becoming a widespread approach to estimate subsurface shear wave velocity profiles. However, the common restriction to vertical component wavefield data does not allow investigations of Love wave dispersion and the partitioning between Rayleigh and Love waves. In this study we extend the modified spatial autocorrelation technique (MSPAC) to three-component analysis (3c-MSPAC). By determination of Love wave dispersion curves, this technique provides additional information for the determination of shear wave velocity–depth profiles. Furthermore, the relative fraction of Rayleigh waves in the total portion of surface waves on the horizontal components is estimated. Tests of the 3c-MSPAC method are presented using synthetic ambient vibration waveform data. Different types of surface waves are simulated as well as different modes. In addition, different spatial distributions of sources are used. We obtain Rayleigh and Love wave dispersion curves for broad frequency bands in agreement with the models used for waveform simulation. The same applies for the relative fraction of Rayleigh waves. Dispersion curves are observed at lower frequencies for Love waves than for Rayleigh waves. While 3c-MSPAC has clear advantages for determination of Love waves velocities, 3c-MSPAC and conventional vertical frequency–wavenumber analysis complement each other in estimating the Rayleigh wave dispersion characteristics. Inversions of the dispersion curves for the shear wave velocity–depth profile show that the use of Love wave velocities confirms the results derived from Rayleigh wave velocities. In the presence of higher mode surface waves, Love waves even can improve results. Application of 3c-MSPAC to ambient vibration data recorded during field measurements (Pulheim, Germany) show dominance of Love waves in the wavefield. Existing shear wave profiles for this site are consistent with models obtained from inversion of Rayleigh and Love wave dispersion curves.