Abstract: A cloud-stripe method is presented for probabilistic seismic demand analysis. This method adopts the strategy commonly used in empirical seismic fragility analysis in order to overcome the limitations of the conventional cloud method in dealing with large amounts of nonlinear time-history data. To illustrate the presented method, 100 real ground motion records are selected as inputs. A total of 23 RC frame structures are used as study cases to perform analyses using the cloud method. Based on the cloud results of all the structures, the probabilistic seismic demand model for the group buildings is derived using the cloud-stripe method, and the corresponding fragility results are calculated. Results show that the cloud-stripe method can effectively handle a large amount of nonlinear time-history analysis data and can develop a reasonable probabilistic seismic demand model. Conditioned on the intensity scope of the considered ground motion records, the group buildings have a limited possibility of occurring severe or complete damage, while their damage outcomes are dominated by slight and moderate damage.