This study was carried out to research and develop a shallow green rooftop system which would require low maintenance and therefore could be used for existing rooftops. To achieve these goals, the conceptual model was induced by past studies and the experimental systems were deduced from the conceptual model. On the growth of Sedum sarmentosum grown in these rooftop systems, the effects of artificial substrate type, soil depth, and drainage type were investigated from 3 April to 11 October 2002. Artificial substrates were an alone type and a blending type. The alone type was an artificial substrate formulated by blending crushed porous glass with bark(v/v, 6:4). The blending type was formulated by blending the alone type with loam(v/v, 1:1). Soil depths were 5cm, loom, and 15cm. Drainage types were a reservoir-drainage type and a drainage type. The reservoir-drainage type could keep water and drain excessive water at the same time. The drainage type could drain excessive water but could not keep water. Covering area, total fresh and dry weight, visual quality, and water content per 1g dry matter were measured. All the variables were analyzed by correlation analysis and factor analysis. The results of the study are summarized as follows. The growth increment was higher in the blending type than in the alone type, the highest in loom soil depth and higher in the reservoir-drainage type than in the drainage type. The growth quality was higher in the blending type than in the alone type, the highest in l0cm soil depth, and higher in the drainage type than in the reservoir-drainage type. In consideration of the permissible load on the existing rooftops and the effects of the treatments on the growth increment and quality, the system should adopt the blending type in artificial substrate types, 5~10cm in soil depths, and the drainage type in drainage types. This system will be well-suited to the growth of Sedum sarmentosum, and when the artificial substrate was in field capacity, the weight will be 75~115kg/$m^2$.