Biodiversity can insure ecosystems against declines in their functioning by increasing the mean level of ecosystem processes and decreasing the spatial or temporal variance of these processes. On this basis, mixing tree species is expected to be an effective management strategy to reduce the risk of planting failure in young plantations. We examined the effects of biodiversity insurance on sapling survival in three tree diversity experiments across Belgium. Based on the survival scoring of 89 254 saplings, planted in 126 plots with up to four-species mixtures, we tested two hypotheses: (i) variability in plot-level survival is lower for mixtures compared to monocultures due to compensation among the species (i.e. buffering effect) and (ii) mean survival is higher due to facilitation (i.e. performance-enhancing effect). Variation in plot-level survival decreased strongly with diversity, indicating a buffering effect. The risk of severe planting failure was reduced in mixtures because species exhibit different survival rates; therefore, mixing ensures that not all trees in the plantation are equally susceptible to environmental stressors. In contrast, the mean plot-level survival did not increase with diversity, and thus, an overall performance-enhancing effect was lacking. However, species-level analyses did show performance-enhancing effects, where some species profited from mixing while others did not. Synthesis and applications. We conclude that biodiversity through mixing tree species insures young experimental plantations against planting failure and is therefore highly recommended as a planting management strategy. The risk of large mortality gaps is reduced if tree plantation saplings are mixed at the scale of individual trees or small cells of trees.