The avaliable data on equilibrium fluctuations in DNA are critically analyzed. Fluctuations of these types are considered: bending and torsional fluctuations and fluctuational opening of base pairs. The latest experimental and theoretical work has made it possible to estimate the bending and torsional stiffnesses of the double helix. Either of these stiffnesses has been determined by at least two different methods that have led to the same results. The stiffness of the double helix is definitely shown to be independent of the concentration of counterions when it exceeds 10(-2) M Na+. As to the fluctuational opening, the current literature is controversial and different methods lead to contradictory results. The method based on DNA unwinding by formaldehyde is considered in some detail. Strong evidence is presented in favour of the modification of base pairs without their complete opening, through an "outside" reaction. Incorporation of this modification route into the theoretical model of DNA unwinding eliminates the last descrepancies between theory and experiment. The successful accounting for all details of the process of DNA unwinding by formaldehyde is argued to prove the adequacy of the helix-coil transition theory as a basis for describing the process of base-pairs opening. The conclusion of McGhee and von Hippel, who obtained a highly overestimated value of the base-pair opening probability proceeding from the formaldehyde kinetics, is shown to be based on an erroneous assumption. The NMR data of Early et al. are briefly discussed. They strongly support the author's concept. A critical comment is made concerning the interpretation of the hydrogen exchange data by Mandel, Kallenbach and Englander. The highly overestimated value of the probability of base-pair opening claimed by these authors is most probably based on unjustified assumptions.