Seismic damage assessment is dealt with in terms of fragility curves. Several approaches for their evaluation, including first-order reliability and a fuzzy random method, are reviewed. These methods are applied to a deterministic infilled reinforced-concrete frame, and the resulting fragility curves compared. Aim is to appraise empirically any difference. Nonstructural damage of the infill is considered. It is assumed to be correlated with the peak interstory drift ratio and dependent on the peak ground acceleration. It is concluded that randomness of the capacity is essential. If a damage state is associated with a deterministic drift range, then fragility steeply increases with the peak ground acceleration, and it is quite overestimated. Comparatively, any other difference among the considered methods results to be minor.