Direct and indirect estimates of turbulence parameters at altitudes 80-120 km are reviewed. It is shown that there are contradictions in data on absolute value, shape, seasonal variations etc. of eddy diffusion coefficient Kt revealed or accepted by different authors. The strongest contradiction is in the views on the seasonal variations of turbulence. Possible explanation may be connected with the correct account for mean motions. Data on turbopause height obtained from neutral composition data (Ar, N2) are discussed. The revealed reverse connection of this height with the temperature at 120 km is discussed in terms of Kt dependence on temperature gradient.