In the current context of anthropogenic global warming, one of the purposes of dendrogeomorphic analyses is to provide long and continuous chronologies of mass movements, so as to detect potential trends or shift related to increasing temperatures. However, on documented slopes, the comparison between historical archives and tree-ring records suggests that dendrogeomorphic reconstructions systematically underestimate the natural activity of the process under investigation. In the specific case of snow avalanches, underestimation generally amounts to 50% and the main causes generally given for this difference are related to the magnitude of past events. In this study, we hypothesize that the morphometric characteristics of avalanche paths and their forest cover could have significant impacts on the length and reliability of tree-ring reconstructions. In order to test this hypothesis, we selected four adjacent, albeit differently structured, avalanche paths from the Queyras massif (French Alps), with the aim to compare their potential for a continuous reconstruction of past avalanche activity. On the most active avalanche paths characterized by high-altitude release areas (covered only by shrubby vegetation), tree-ring reconstructions do not exceed one century in length, with recurrence intervals of high magnitude events >25 years. By contrast, on forested couloirs where lower slopes and forest coverage up to the release areas limits the intensity of events, the frequency of reconstructed snow avalanches is 2.5 times higher, the reconstructions span longer periods and the convergence rate with historical archives attest to the reliability of the dendrogeomorphic approach. These results suggest that a careful selection of couloirs is essential and that priority should be given to forested sites as (i) they allow for exhaustive and (ii) reliable reconstructions over (iii) long periods of time.