While rising global temperatures have altered global drought risk and are projected to continue to change large-scale hydroclimate, it has proved difficult to detect the influence of warming on drought-relevant variables at regional scales. In addition to the inherent difficulty in identifying signals in noisy data, detection and attribution studies generally rely on general circulation models, which may fail to accurately capture the characteristics of naturally forced and internal hydroclimate variability. Here, we use a long tree-ring based paleoclimate record of drought to estimate pre-industrial variability in the Palmer Drought Severity Index (PDSI), a commonly used metric of drought risk. Using a Bayesian framework, we estimate the temporal and spatial characteristics of hydroclimate variability prior to 1850. We assess whether observed twenty-first century PDSI is compatible with this pre-industrial variability or is better explained by a forced response that depends on global mean temperature. Our ressults suggest that global warming likely contributed to dry PDSI in Eastern Europe, the Mediterranean, and Arctic Russia and to wet PDSI in Northern Europe, East-central Asia, and Tibet.