Rapid ecosystem transitions and adverse effects on ecosystem services as responses to combined climate and human impacts are of major concern. Yet few long-term (i.e. >60 years) quantitative observational time series exist, particularly for ecosystems that have a long history of human intervention. Here, we combine three major environmental pressures (land use, nutrients and erosion) with quantitative summer and winter climate reconstructions and climate model simulations to explore the system dynamics, resilience and the role of disturbance regimes in varved eutrophic Lake Żabińskie (NE Poland) since AD 1000. The comparison between these independent sources of information allows us to establish the coherence and points of disagreements between such data sets. We find that climate reconstructions capture noticeably natural forced climate variability, while internal variability is the dominant source of variability during most parts of the last millennium at the regional scale, precisely at which climate models seem to underestimate forced variability. Using different multivariate analyses and change point detection techniques, we identify ecosystem changes through time and shifts between rather stable states and highly variable ones. Prior to AD 1600, the lake ecosystem was characterised by high stability and resilience against observed natural climate variability. During this period, the anthropogenic fingerprint was small; the lake ecosystem was buffered against the combined human and natural disturbance. In contrast, lake–ecosystem conditions started to fluctuate across a broad range of states after AD 1600. The period AD 1745–1886 represents the phase with the strongest human disturbance of the catchment–lake ecosystem. During that time, the range of natural climate variability did not increase. Analyses of the frequency of change points in the multi-proxy data set suggest that the last 400 years were highly variable and increased vulnerability of the ecosystem to the anthropogenic disturbances. This led to significant rapid ecosystem transformations.