Tree plantings have the potential to increase species diversity and sequester carbon, yet planting failure and early mortality pose significant barriers to their success. Biodiversity‐ecosystem function theory suggests that diverse tree plantings could improve survival outcomes through either the portfolio or facilitation effect, yet there remain few tests of this hypothesis. Here, we use a large‐scale tree‐diversity experiment (BiodiversiTREE), with monitoring of nearly 8,000 individual trees to test whether (1) tree species diversity increases survival rates, (2) tree diversity stabilizes the risk of planting failure, and/or (3) diversity effects are important relative to other common drivers of seedling mortality (e.g. herbivory and soil moisture). We found that only species identity significantly impacted the likelihood of survival, not plant functional diversity nor plot species richness nor phylogenetic diversity. There were minor effects of elevation and soil moisture on survival, but both explained a very small amount of variation in the data (r2marg ≤ 0.011). Higher tree diversity did, however, strongly reduce variation in survival across plots, with nearly 2‐fold higher coefficients of variation in monocultures (30.4%, 28.4–32.6% 95% bootstrapped confidence interval) compared to 4‐ (16.3%, 13.8–18.7%) and 12‐species plots (12.8%, 10.8–14.7%). Ultimately, our results suggest that employing diverse species can lower the risk of planting failure (i.e. the portfolio effect), but that species selection still plays a large role in early establishment.