The assessment of the surface roughness effect in fatigue design is a complex task. The fatigue strength of cyclic loaded mechanical components is determined by local geometric notches and the manufacturing technology specific material resistance. Thus, macroscopic surface features, as inherited by a cast surface, lead to stress concentration due to notch effect and lowers the component's fatigue strength decisively. Hence, this research work focuses on the development of an engineering feasible fatigue strength assessment concept of heterogeneously textured sand-cast surfaces. Therefore, crankcases of the conventional aluminium alloy EN AC-46200 in T6 and HIP+T6 heat treatment condition are studied in terms of metallographic characterisation, quasi-static properties, fatigue strength and characteristic surface texture features. For specimens with cast surface, basically two failure mechanisms are observed. Fatigue cracks either initiate only at a surface texture based notch, or at a combinatoric defect case involving a surface layer pore that interacts with the cast surface notch. At first, to assess the notch effect resulting from the cast surface, a common stress concentration factor is modified. Notch depth and notch root radius are thereby locally characterised by means of an innovative sub-area based approach. The developed model uses the areal maximum pit height roughness parameter Sv as notch depth representative qualifier and the loading-direction independent equivalent notch root radius ρ based on the mean curvature H, thus reducing directional effects. Moreover, a statistical characterisation provides additional model parameters and contributes to the evaluation concept's improvement regarding a probabilistic cast surface fatigue strength estimation. At second, the cast surface layer fatigue strength assessment is extended with regard to the defect afflicted surface layer, and features a combinatoric approach of porosity and surface notch effects by means of an interaction coefficient, computed by a neural network. The overall concept is proven valid for the investigated aluminium sand-cast surface layers with a conservative fatigue strength estimation in the range of six to nine percent. This features an engineering feasible assessment tool for heterogeneous sand-cast surface textures with theoretical applicability on alike manufacturing processes such as additively manufactured surface textures.