Since the 3+1 neutrino models with one light sterile neutrino turn out to be not very effective, at least two light sterile neutrinos may be needed to reconcile the solar and atmospheric neutrino experiments with the LSND result,  if this is confirmed by the ongoing MiniBooNE experiment (and when the CPT invariance is assumed to hold for neutrino oscillations). We present an attractive 3+2 neutrino model, where two light sterile neutrinos mix maximally with each other, in analogy to the observed maximal mixing of muon and tauon active neutrinos. But, while the mixing of \nu _e and (\nu _\mu - \nu _\tau )/\sqrt 2 is observed as large (though not maximal), the mixing of \nu _e with the corresponding combination of two light sterile neutrinos is expected to be only moderate because of the reported smallness of LSND oscillation amplitude. The presented model turns out, however, not to be  more effective in explaining the hypothetic LSND result than the simplest 3+1 neutrino model. On the other hand, in the considered 3+2 model, the deviations from conventional oscillations of three active neutrinos appear to be  minimal within a larger class of 3+2 models.

PACS numbers: 12.15.Ff, 14.60.Pq, 12.15.Hh