Influence of compressive strength scatter on the fatigue resistance of concrete under compression. Experimentally determined load cycles to failure of concrete specimens under compression generally show a large scatter. The main cause is assumed to be the scatter of the real compressive strength, since the upper and lower stresses applied in the fatigue test are always related to the average compressive strength. Accordingly, specimens that really have higher compressive strengths fail at higher numbers of load cycles than specimens with lower real compressive strengths. In order to verify this assumption, the following article presents a concept for the stochastic consideration of compressive strength scatter in the derivation of S-N curves from tests. The concept is developed on the basis of extensive fatigue investigations on concrete of strength classes C100/115 and C80/95 for wind turbine towers and leads to almost smooth, scatter-corrected S-N curves. This indicates that the original scatter inherent in the S-N curves can be explained by the scatter of the compressive strength. The application of this concept is limited to the boundary conditions under which the experimental investigations were carried out. In supplementary investigations, the prevailing humidity, temperature and frequency conditions in a wind turbine tower during operation are investigated and compared with the conditions under which the laboratory results were obtained. Finally, a design concept is formulated for the investigated concretes, which recommends the application of the S-N curves according to fib Model Code 2010 including an adapted formulation for the design compressive strength at fatigue loads f_cd,fat.