The attribution of past glacier length fluctuations to changes in climate requires characterizing glacier mass-balance variability. Observational records, which are relatively short, are consistent with random fluctuations uncorrelated in time, plus an anthropogenic trend. However, longer records of other climate variables suggest that, in fact, there is a degree of temporal persistence associated with internal (i.e. unforced) climate variability, and that it varies with location and climate. Therefore, it is likely that persistence does exist for mass balance, but records are too short to confirm its presence, or establish its magnitude, with conventional statistical tests. Extending the previous work, we explore the impact of potential climatic persistence on glacier length fluctuations. We use a numerical model and a newly developed analytical model to establish that persistence, even of a degree so small as to be effectively undetectable in the longest mass-balance records, can significantly enhance the resulting glacier length fluctuations. This has a big impact on glacier-excursion probabilities: what was an extremely unlikely event (<1%) can become virtually certain (>99%), when persistence is incorporated. Since the actual degree of climatic persistence that applies to any given glacier is hard to establish, these results complicate the attribution of past glacier changes.