Man-made vitreous fibers (MMVF) comprise diverse materials for thermal and acoustic insulation, including stone wool. Depending on dimension, durability, and dose, MMVF might induce adverse health effects. Therefore, early predictive in vitro (geno)toxicity screening of new MMVF is highly desired to ensure safety for exposed workers and consumers. Here, we investigated, as a starting point, critical in vitro screening determinants and pitfalls using primary rat alveolar macrophages (AM) and normal rat mesothelial cells (NRM2). A stone wool fiber (RIF56008) served as an exemplary MMVF (fibrous vs. ground to estimate impact of fiber shape) and long amosite (asbestos) as insoluble fiber reference. Materials were comprehensively characterized, and in vivo-relevant in vitro concentrations defined, based on different approaches (low to supposed overload: 0.5, 5 and 50 µg/cm2). After 4–48 h of incubation, certain readouts were analyzed and material uptake was investigated by light and fluorescence-coupled darkfield microscopy. DNA-strand break induction was not morphology-dependent and nearly absent in both cell types. However, NRM2 demonstrated material-, morphology- and concentration-dependent membrane damage, CINC-1 release, reduction in cell count, and induction of binucleated cells (asbestos > RIF56008 > RIF56008 ground). In contrast to NRM2, asbestos was nearly inactive in AM, with CINC-1 release solely induced by RIF56008. In conclusion, to define an MMVF-adapted, predictive in vitro (geno)toxicity screening tool, references, endpoints, and concentrations should be carefully chosen, based on in vivo relevance, and sensitivity and specificity of the chosen cell model. Next, further endpoints should be evaluated, ideally with validation by in vivo data regarding their predictivity.