Comparison of accident-hazard potentials associated with neutron-activation products in fusion reactors of various designs and structural materials suffers from a number of shortcomings in the readily available hazard-index data. Neither inventories of curies nor biological hazard potentials (BHPs) are satisfactory indices of hazard even if consistently computed, and between-study inconsistencies in neutronics packages and BHP calculations further obscure the meaning of comparisons based on these measures. We present here the results of internally consistent calculations of radioactive inventories, BHPs, and off-site dose potentials associated with the first walls of nine reactor-design/first-wall-material combinations. A recent mirror-reactor design reduces off-site dose potentials by a factor of 2 compared to a muchstudied early tokamak, for a given first-wall material. Holding design fixed, HT-9 ferritic steel offers a factor of 2 reduction in dose potential compared to Type 316 stainless steel. By the dose-potential measure, molybdenum is the worst of the materials investigated and silicon carbide is by jar the best. Hazards in realizable accidents depend not only on the hypothetical dose potentials, as calculated here, but also on the actual release fractions of first-wall (or other activated) material. Review of the theoretical and experimental evidence bearing on release fractions suggests that, for most candidate materials, high release fractions from designs containing liquid lithium cannot yet be convincingly ruled out.