We estimate the uncertainty of satellite‐retrieved Arctic sea‐ice thickness, sea‐ice volume, and their trends stemming from the lack of reliable snow‐thickness observations. To do so, we simulate a Cryosat2‐type ice‐thickness retrieval in an ocean‐model simulation forced by atmospheric reanalysis, pretending that only freeboard is known as model output. We then convert freeboard to sea‐ice thickness using different snow climatologies and compare the resulting sea‐ice thickness retrievals to each other and to the real sea‐ice thickness of the reanalysis‐forced simulation. We find that different snow climatologies cause significant differences in the obtained ice thickness and ice volume. In addition, we show that Arctic ice‐volume trends derived from ice‐thickness retrievals using any snow‐depth climatology are highly unreliable because the estimated trend in ice volume can strongly be influenced by the neglected interannual variability in snow volume.