In the design of nanoparticles that can target disease tissue in vivo, parameters such as targeting ligand density, type of target receptor, and nanoparticle shape can play an important role in determining the extent of accumulation. Herein, a systematic study of these parameters for the targeting of mouse xenograft tumors is performed using superparamagnetic iron oxide as a model nanoparticle system. The type of targeting peptide (recognizing cell surface versus extracellular matrix), the surface coverage of the peptide, its attachment chemistry, and the shape of the nanomaterial [elongated (nanoworm, NW) versus spherical (nanosphere, NS)] are varied. Nanoparticle circulation times and in vivo tumor-targeting efficiencies are quantified in two xenograft models of human tumors (MDA-MB-435 human carcinoma and HT1080 human fibrosarcoma). It is found that the in vivo tumor-targeting ability of the NW is superior to that of the NS, that the smaller, neutral CREKA targeting group is more effective than the larger, positively charged F3 molecule, that a maximum in tumor-targeting efficiency and blood half-life is observed with ≈60 CREKA peptides per NW for either the HT1080 or the MDA-MB-435 tumor types, and that incorporation of a 5-kDa polyethylene glycol linker improves targeting to both tumor types relative to a short linker. It is concluded that the blood half-life of a targeting molecule–nanomaterial ensemble is a key consideration when selecting the appropriate ligand and nanoparticle chemistry for tumor targeting.