There are two main classes of metallic nanoparticles: solid and hollow. Each type can be synthesized in different shapes and structures. Practical use of these nanoparticles depends on the properties they acquire on the nanoscale. Plasmonic nanoparticles of silver and gold are the most studied, with applications in the fields of sensing, medicine, photonics, and catalysis. In this Account, we review our group's work to understand the catalytic properties of metallic nanoparticles of different shapes. Our group was the first to synthesize colloidal metallic nanoparticles of different shapes and compare their catalytic activity in solution. We found that the most active among these were metallic nanoparticles having sharp edges, sharp corners, or rough surfaces. Thus, tetrahedral platinum nanoparticles are more active than spheres. We proposed this happens because sharper, rougher particles have more valency-unsatisfied surface atoms (i.e., atoms that do not have the complete number of bonds that they can chemically accommodate) to act as active sites than smoother nanoparticles. We have not yet resolved whether these catalytically active atoms act as catalytic centers on the surface of the nanoparticle (i.e., heterogeneous catalysis) or are dissolved by the solvent and perform the catalysis in solution (i.e., homogenous catalysis). The answer is probably that it depends on the system studied. In the past few years, the galvanic replacement technique has allowed synthesis of hollow metallic nanoparticles, often called nanocages, including some with nested shells. Nanocage catalysts show strong catalytic activity. We describe several catalytic experiments that suggest the reactions occurred within the cage of the hollow nanocatalysts: (1) We synthesized two types of hollow nanocages with double shells, one with platinum around palladium and the other with palladium around platinum, and two single-shelled nanocages, one made of pure platinum and the other made of pure palladium. The kinetic parameters of each double-shelled catalyst were comparable to those of the single-shelled nanocage of the same metal as the inside shell, which suggests the reactions are taking place inside the cavity. (2) In the second set of experiments, we used double-shelled, hollow nanoparticles with a plasmonic outer gold surface and a non-plasmonic inner catalytic layer of platinum as catalysts. As the reaction proceeded and the dielectric function of the interior gold cavity changed, the plasmonic band of the interior gold shell shifted. This strongly suggested that the reaction had taken place in the nanocage. (3) Finally, we placed a catalyst on the inside walls of hollow nanocages and monitored the corresponding reaction over time. The reaction rate depended on the size and number of holes in the walls of the nanoparticles, strongly suggesting the confinement effect of a nanoreactor.