Nanomaterials (NMs) can be produced in plenty of variants posing several challenges for NM hazard and risk assessment. Metabolomic profiling of NM-treated cells and tissues allows for insights into underlying Mode-of-Action (MoA) and offers several advantages in this context. It supports the description of Adverse Outcome Pathways (AOPs) and, therefore, tailored AOP-based hazard testing strategies. Moreover, it bears great potential for biomarker discovery supporting toxicity prediction. Here, we applied metabolomics profiling to cells treated with four well-selected SiO2 variants, differing in structure, size and surface charge. TiO2 NM-105 served as a benchmark. Responses were studied in vitro in rat lung epithelial cells (RLE-6TN) and alveolar macrophages (NR8383) and compared to in vivo responses in rat lung tissues obtained from in vivo instillation and short-term inhalation studies (STIS). Time- and concentration-dependent changes were observed in both in vitro models but with cell-type specific responses. Overall, the levels of lipids and biogenic amines (BAs) tended to increase in epithelial cells but decreased in macrophages. Many identified metabolites like Met-SO, hydroxy-Pro and spermidine were related to oxidative stress, indicating that oxidative stress contributes to the MoA for the selected NMs. Several biomarker candidates such as Asp, Asn, Ser, Pro, spermidine, putrescine and LysoPCaC16:1 were identified in vitro and verified in vivo. In this study, we successfully applied a metabolomics workflow for in vitro and in vivo samples, which proved to be well suited to identify potential biomarkers, to gain insights into NM structure–activity relationship and into the underlying MoA.