We show that the experimentally observed behavior of thermal conductivity of dielectric glasses over a wide temperature range can be explained by a combination of two scattering processes. The first one comes from the phonon scattering due to biaxial dipoles of wedge disclinations while the second one is the Rayleigh type scattering. The results obtained support the cluster picture suggested earlier for glassy materials.Physics of glassy systems has been attracting a considerable interest for a long period. Since the discovery of the anomalous low-temperature behavior of amorphous dielectrics [1] there were many attempts to explain this unusual phenomenon. And yet, in 1986 Freeman and Anderson [2] reviewing both numerous empirical results and theoretical models made a rather drastic conclusion that the thermal conductivity of amorphous dielectrics "is not understood in any temperature range". While at first glance this conclusion looks too pessimistic, it reflects an actual situation existed in glassy physics at that time. It has been found experimentally that the low-temperature anomaly in thermal conductivity, κ, observed by Zeller and Pohl for some noncrystalline solids [1] is universal. In particular, a T 2 dependence of κ below 1K together with the following characteristic plateau region have been found in the most of glassy materials (see, e.g., review [3]) as well as in some quasicrystalline alloys [4]. This finding allows to assume that a realistic theoretical model for description of the thermal conductivity should capture the very essence of these materials, their microstructure. Thus the conclusion of Freeman and Anderson actually suggests that this problem still remains to be solved.The most widely used model which explains glassy physics below 1K has been proposed by Phillips [5] and Anderson, Halperin, and Varma [6]. In this temperature region, thermal transport takes place via propagating acoustic modes. Thus, κ depends on the scattering of the phonons by the structure. Within the model [5,6], the principal scatterers were proposed to be the tunneling states (TS) in the glass. However, in spite of the success achieved in interpreting experimental data, some important questions were left unanswered. The main of them concern the nature of TS and the source of their universality (see, e.g., discussion in [3,7]).Since 1972 there have been many attempts to explain the experimental data in terms of propagating thermal phonons scattered by TS [3,[8][9][10][11] (in different modifications), density fluctuations (Rayleigh scattering) [1,3,10,[12][13][14], and the like. Some papers invoked nonpropagating vibrational modes as well [3]. The most successful approach includes a combination of TS and the Rayleigh scattering [10]. Notice that there were attempts to go beyond the TS approximation (see Ref.[15] and the references therein) as well as to do without TS at all. For example, an interesting phenomenological approach was discussed in Ref. [3] where a combination of the structure scattering together with a...