The Casimir forces induced by fluctuations of electromagnetic field have been measured with a high precision at distances above 60 nm. Although at shorter separations the magnitude of the force increases, it is more complicated to measure the forces at distances below 50 nm. We review the problems that appear in this range of distances and their possible solutions. The first problem is related to the pull-in instability that occurs when two surfaces get too close to each other. As a particular manifestation of this problem, the spontaneous formation of the capillary bridges at distances ∼10 nm is discussed. As an alternative, we discuss the method of adhered cantilever, which does not suffer from the pull-in instability. The second problem is related to the roughness of interacting surfaces that gives a significant deviation of the force from the expected scaling with the distance. We explain how the deviation can be related to the contribution of high asperities to the force. Characterization of the deposited rough films is also covered with a special emphasis on the excessive number of high asperities for the films deposited under nonequilibrium conditions. The third problem is related to the poor precision in the determination of the absolute distance between the bodies that results in a large total error in the force. We discuss the methods to determine the distance upon contact and cover a proposition to improve this precision in the method of adhered cantilever.