Steel frames not only have the advantages of being light weight，and high strength but also are corrosion resistant and fireproof，so steel is widely used for industrial construction.Unfortunately，research on corrosion mechanisms and measures of corrosion protection were mainly aimed at the timber level and based on experiments.Due to differences in test conditions，test methods and other factors，conclusions from these experiments were highly variable and empirical formulas were difficult to utilize.In addition，research aimed at component-level corrosion was also not yet clear，so uniform corrosion is often assumed for convenience.For example，some simply reported weakened cross-section members and considered material deterioration to analyze structural performance. In this paper，we consider the seismic performance of overall structure in different service times.Material deterioration and non-uniform corrosion were accounted for in column and beam cross-sections.Steel corrosion caused the deterioration of seismic performance and stiffness for structures.When steel frame structures survive earthquakes，it does not guarantee that the structures will survive one that reaches the design capacity at a later service time.Therefore，when evaluating the seismic performance of steel frame structures in different service times，we must account for the increased seismic risks caused by steel corrosion over time.Changes in seismic performance of structures over different service times were examined in this paper. The deterioration of elastic modulus and steel yield strength were evaluated using existing empirical formulas. At the same time，the thickness variation of the column and beam cross sections in steel frame structures was accounted for by using findings from related research.Different methods including IDA，pushover，and Monte Carlo were combined together to estimate values for different limit states in a probabilistic seismic capacity model. In this paper，the maximum drift angle was taken as the damage index. The statistical parameters of the probabilistic seismic capacity model were estimated through pushover analysis，while the relationship of probabilistic seismic demand was obtained by IDA analysis for random structure samples and different service times.Seismic fragility surfaces of steel-frame structures were obtained in different limit state and different service time （0 year，15 years， 30 years，45 years，and 60 years）. Finally, a case study of a 15-floor steel-frame structure was adopted to illustrate the change in structure fragility that may occur as a result of corrosion. Seismo-structure software and finite-element models were used to test different service times of the structure. Pushover and IDA analysis were then used to calculate values of different limit states in a probabilistic seismic capacity model and the relationships of probabilistic seismic demand models，respectively，over different service times. These achievements would provide theoretical support for the life-cycle design of steel frame structures，operation and management，and especially seismic hazard assessment of existing steel frame structures.