In this study we provide guidance on the biologically most relevant dose metric for pulmonary toxicity of biopersistent, spherical nanoparticles (NPs). A retrospective analysis of nine in vivo studies on particle-induced, acute pulmonary toxicity in animal models (mouse, rat) was performed encompassing five different types of nanomaterials (polystyrene, titanium dioxide, carbonaceous materials, transition metal oxides (Co, Ni, Zn) and hydrothermally synthesized α-quartz) with a wide range of primary particle diameters (9-535nm) and mass-specific BET surface areas (6-800m2/g). The acute influx of polymorphonuclear neutrophils (PMNs) into the lungs after intratracheal instillation of NPs was chosen as a toxicological endpoint for acute lung inflammation. The allometrically scaled toxicological data were investigated with respect to various dose metrics, namely (primary) particle number, joint length, BET and geometric surface area, volume and mass.Surface area is identified as the biologically most relevant dose metric for spherical NPs explaining about 80% of the observed variability in acute pulmonary toxicity (R 2=0.77). None of the other dose metrics explains more than 50% of the observed variability in pulmonary inflammation. Moreover, using surface area as the dose metric allows identification of material-based toxicity classes independent of particle size. Typical materials without intrinsic toxicity - here referred to as low-solubility, low-toxicity (LSLT) materials - show low surface-specific toxicity with an EC50 dose of 175m2/g-lung (geometric mean; σ g =2.2), where EC50 represents the dose inducing 50% of the maximum effect (here 30% PMN). In contrast, transition metal oxides (here Co, Ni, Zn) - materials known for their intrinsic toxicity - display a 12-fold enhanced surface-specific toxicity compared to LSLT particles (EC50=15m2/g-lung).This analysis implies that surface-related modes of action are driving acute pulmonary toxicity for the types of NPs investigated here. The relevance of other dose metrics such as number and volume is acknowledged in the context of different modes of action, namely shape-induced toxicity (fiber paradigm) and extremely high particle lung burden (overload conditions), respectively. So which dose metric should be monitored by aerosol scientists involved in health related aerosol exposure measurements? The short answer is - all of them (except length), but there is a strong preference towards surface area.