This work investigates sulfate attack in complex sulfate environments by exposing different binder types to various sulfate solutions and comparing predicted phase and volume changes with experimental data. The most important aspects of this work can be grouped in three topics: The comparison of the predicted volume increase with the experimentally observed length changes. This part of the work shows that volume increase cannot be linked directly to the observed expansions. Additionally, the volume increase model does not provide an explanation for the driving force for the space filling to overcome mechanical constraints. A more plausible explanation of expansion lies in the theory of crystallization pressure, in which crystals forming from a supersaturated solution seem to be a more applicable explanation for the deterioration differences. However, this is difficult to verify directly as it is impossible to measure the solution concentration or individual crystals exert pressure on their surroundings. It is observed that expansion occurs in systems where thermodynamic modelling predicts the co-existence of ettringite with gypsum. In such a case, if monosulfate and gypsum are both present locally, the solution can be highly supersaturated with respect to ettringite within the exposed materials. An increase of sulfate contents in the C-S-H phase has been observed for these cases, indicating increased sulfate concentrations in the pore solution while providing the necessary confinement for the crystals in intimate phase mixtures to exceed pressure. The presence of bicarbonate ions in the sulfate solution reduces the expansion significantly for the CEM I and CEM III/B binders, and reduced the sulfate uptake in the phase assemblage. The CEM III/B mortars showed a highly leached zone at the surface in which also calcite was observed, this is attributed to the destabilisation of ettringite in the presence of high concentrations of bicarbonate ions. The microstructural characterization combined with the information from thermodynamic modelling suggests that conditions of high supersaturation with respect to ettringite are unlikely to occur in samples exposed to solutions containing bicarbonate ions. Degradation of mortars exposed to complex magnesium containing sulfate solutions showed that the presence of sodium, potassium and calcium in a magnesium solution reduce the deterioration symptoms significantly for the CEM I and CEM III/B mortars. The experimental observations suggest that sulfate attack in natural environments is not only less severe, due to reduced sulfate concentrations in the field, but are likely to be affected by the presence of bicarbonate anions and the common occurrence of different cations. Both of these aspects reduce the deterioration significantly and can be assumed to occur in most natural waters.