Release of neutron-activation products in severe hypothetical fusion-reactor accidents may constitute a larger health hazard than that of the tritium released at the same time. Significant escape of activation products could result from lithium fires hot enough to melt and partly vaporize activated first-wall materials, or from other accident sequences that bring air into contact with activated structure hot enough to cause the formation of volatile metal oxides. Analysis of three combinations of structural materials and severe accident scenarios has been undertaken for an early conceptual tokamak reactor, using a simple consequence model based on that of the Nuclear Regulatory Commission's Reactor Safety Study (the Rasmussen report) to determine conceivable radiation doses near the plant boundary. (No attempt was made to estimate probabilities for such severe events.) In the cases of stainless-steel and molybdenum structures subject to massive lithium fires, the boundary doses far exceed those that would be produced by release of the entire plant inventory of tritium and are comparable to the doses similarly calculated for “worst case” light water reactor accidents. The case of niobium fusion-reactor structure is more favorable. These results, based on an early fusion-reactor design not optimized with respect to safety characteristics, may well portray a worst case picture of fusion accident consequences. They suggest, however, that the large potential safety advantages of fusion compared to fission are not necessarily inherent for all designs and choices of materials, and they motivate attention to the several available strategies for greatly reducing the potential for activation-product release from fusion reactors.