Abstract. When making multi-proxy reconstructions over Holocene-long periods, an argument is presented that the elevation of the sites used and/or their elevation history must be taken into account before their proxy records (of temperature or precipitation) are included in the reconstruction. It is shown that to ignore elevation, results in first order errors in the reconstruction, especially in regions under (or close to) the degrading ice sheet. Also it is argued that when assessing the signature of a given putative global event (like the 4.2 ka event), one must allow for there being a complex signature wrt. location, elevation and time lagged variables. Three specific examples are used to illustrate these points.