Abstract It has been suggested that maximum latewood density (MXD) should be used instead of tree-ring width (TRW) data to reconstruct post-volcanic cooling effects. A thorough assessment of high frequency signals and potentially differing memory effects in long MXD and TRW chronologies, in response to large volcanic eruptions, is still missing, however. We here present a compilation of MXD and TRW chronologies from 11 sites in the Northern Hemisphere, covering the past 750+ years, and containing significant June–August temperature signals. Basic assessment of the data using Superposed Epoch Analysis reveals a temporally extended response in TRW, by 2–3 years, to large volcanic eruptions, though post-volcanic cooling patterns vary considerably within the Northern Hemisphere network. Comparison with instrumental temperature data demonstrates the TRW chronologies underestimate cold conditions associated with large volcanic eruptions, a bias that is mitigated in the MXD data. While species composition (pine, spruce, larch) has no detectable influence on the cooling patterns, trees from high latitude sites (>60°N) indicate a stronger and delayed (1–2 years) response to large eruptions, compared to the lower latitude sites (<60°N). These basic findings caution against using TRW data for quantitatively estimating post-volcanic cooling and for comparison against the simulated climate effects of volcanic eruptions in models.