Nanoparticles (NP) have a tendency to agglomerate after dispersion in physiological media, which can be prevented by the addition of serum. This may however result in modification of the toxic potential of particles due to the formation of protein corona. Our study aimed to analyze the role of serum that is added to improve the dispersion of 10 nm TiO2 NPs on in vitro and in vivo effects following the exposure via the respiratory route. We characterized NP size, surface charge, sedimentation rate, the presence of protein corona and the oxidant-generating capacity after NP dispersion in the presence/absence of serum. The effect of serum on NP internalization, cytotoxicity and pro-inflammatory responses was assessed in a human pulmonary cell line, NCI-H292. Serum in the dispersion medium led to a slower sedimentation, but an enhanced cellular uptake of TiO2 NPs. Despite this greater uptake, the pro-inflammatory response in NCI-H292 cells was lower after serum supplementation (used either as a dispersant or as a cell culture additive), which may be due to a reduced intrinsic oxidative potential of TiO2 NPs. Interestingly, serum could be added 2 h after the NP treatment without affecting the pro-inflammatory response. We also determined the acute pulmonary and hepatic toxicity in vivo 24 h after intratracheal instillation of TiO2 NPs in C57BL/6N mice. The use of serum resulted in an underestimation of the local acute inflammatory response in the lung, while a systemic response on glutathione reduction remained unaffected. In conclusion, serum as a dispersion agent for TiO2 NPs can lead to an underestimation of the acute pro-inflammatory response in vitro and in vivo. To avoid potential unwanted effects of dispersants and medium components, we recommend that the protocol of NM preparation should be thoroughly tested, and reflect as close as possible realistic exposure conditions.