Green roofs are known to mitigate the negative effects of urban consolidation by offering diverse ecosystem functions compared to non-vegetated roofs. However, the support for native biodiversity might be improved by using native plant species. In a mesocosm experiment, we studied the suitability of three commercial green-roof growth substrates for the establishment of 27 native plant species from dry sandy grasslands of northwestern Germany over the course of four years. The substrates were mineral-based, but differed in the layering of organic matter. Total establishment rates reached 44–59% in Year 4, indicating the general suitability of the substrates. During the first weeks after seeding, with light irrigation, the vascular plant cover was greater in the similar substrates Zincolit® Plus (Z) and Zincolit® Plus-Leicht (ZL) with their compost-based organic mulch layers than in the substrate Sedumteppich (ST) with its organic matter evenly admixed with the mineral aggregates. In Years 2 and 3, however, the vascular plant cover was greater in the ST substrate, likely due to the better availability of water and nutrients from the organic matter compared to the dry surface-mulch layer variants Z and ZL. After severe drought events, the decline in plant cover was more pronounced in the ST substrate, likely representing a trade-off between lush growth and a susceptibility to drought. An indicator-species analysis revealed differences in species composition between the ST and Z/ZL substrates. Annual plant species were indicators of the ST substrate. Perennials, such as Thymus pulegioides and Achillea millefolium, were typical of the Z and ZL substrates. In addition to the general suitability of the tested standard substrates for target species establishment, the study indicated that a combination of different layers of substrate components resulted in different vegetation patterns that may have a positive effect on green-roof biodiversity.