A realistic prediction of the concrete fatigue life exposed to high-cycle loading scenarios with variable amplitudes is of utmost importance for a reliable and economically efficient design of civil engineering infrastructure for transport and energy supply. Current design codes estimate the fatigue life under variable amplitudes using the Palmgren–Miner rule, which assumes a linear scaling between lifetimes measured for uniform cyclic loading scenarios. Several experimental series conducted in the past, however, indicate that this assumption is not valid and that it may lead to unsafe design. In this paper, an experimental and theoretical investigations of the fatigue loading sequence effect in normal- and high-strength concrete behavior are presented, which confirm this observation. In particular, a test campaign with 135 cylinder specimens, including three concrete grades and six different loading scenarios has been conducted. Several response characteristics of the fatigue behavior including Wöhler curves, fatigue creep curves and evolving shapes of hysteretic loops have been evaluated. To substantiate the experimental results, a theoretical explanation of the observed sequence effect is formulated based on the assumption, that energy is dissipated uniformly within the volume of a test specimen during subcritical, compressive cyclic loading. Then, superposition of energy dissipation profiles along the lifetime measured for constant amplitudes becomes possible and a theoretical justification of the experimentally observed sequence effect can be provided. Moreover, a reverse sequence effect reported in the literature for bending fatigue of concrete can then be explained by an unevenly distributed energy dissipation over a cracked specimen. Supported by the theoretical consideration, the processed experimental data is used to validate existing fatigue life assessment rules by testing their ability to reflect the load sequence effect.