Extreme Weather Events (EWE), such as late frosts and summer droughts, are increasingly frequent and drive devastating impacts on ecosystems. Non-linear responses are commonly associated with EWE but are poorly captured by traditional replicated experimental designs. Further, it is unclear how thresholds associated with one event type are modulated by other extreme events. We present the first manipulative experiment that explicitly assesses the prevalence of non-additive and non-linear responses to multiple EWE using a crossed-gradient approach. We exposed 450 seedlings of seven tree species that are common in Central European forests (Abies alba, Fagus sylvatica, Larix decidua, Picea abies, Pinus mugo ssp. rotundifolia, Pseudotzuga menziesii and Sorbus torminalis) to late frosts of varying severity (six regimes, -2°C to -11°C) and to droughts of varying lengths (ten regimes, no drought through to 12 weeks’ drought). We measured a variety of ecologically relevant responses: growth, mortality, phenology and biomass allocation. While responses varied across species, EWE near-invariably elicited non-linear responses irrespective of species or response parameter of interest. We found similar prevalence of non-additive EWE interactions. Our experiment highlights the relevance of manipulative approaches that enable researchers to resolve interactive and non-linear effects of EWE and other global environmental change drivers.