Typical structural systems have a high degree of static indeterminacy, but a low degree of redundancy. Abnormal loads of large intensity, but small probability of occurrence, like fires, landslides, accidental impact, and explosions can produce localized damage. With lack of or with insufficient redundancy, localized initial damage may propagate, leading to the progressive collapse of the whole or of a disproportional part of the structure. On the other hand, strengthening the structure to produce alternate load paths (redundancy) has a significant impact in construction costs, with unknown cost-benefit relation. In this talk we address the problem from a risk-optimization perspective, which takes into account the possible system failure states of regular frame structures, as well as the corresponding costs of failure. A cost-benefit analysis is performed, comparing the strengthening cost with the reduction in expected failure costs, given local damage produced by abnormal loads. As a result, we obtain a break-even local damage or hazard probability, for which these costs are the same. For threats and hazards leading to local damage probabilities higher than the break-even value, structural strengthening is cost-effective. We show how the break-even probability changes with relevant problem parameters like strengthening cost, costs of failure, aspect ratio of the buildings, and extend of the strengthening measure. We show how the different failure propagation mechanisms compete for the limited strengthening budget, and we propose optimal design factors for beams and columns of regular frame buildings.