The last decade has seen the refinement of a technique for reconstructing palaeo-ice thicknesses based on using the retained H2O and CO2 content in glassy eruptive deposits to infer quenching pressures and therefore ice thicknesses. The method is here applied to Bláhnúkur, a subglacially erupted rhyolitic edifice in Iceland. A decrease in water content from ∼0.7 wt.% at the base to ∼0.3 wt.% at the top of the edifice suggests that the ice was 400 m thick at the time of the eruption. As Bláhnúkur rises 350 m above the surrounding terrain, this implies that the eruption occurred entirely within ice, which corroborates evidence obtained from earlier lithofacies studies. This paper presents the largest data set (40 samples) so far obtained for the retained volatile contents of deposits from a subglacial eruption. An important consequence is that it enables subtle but significant variations in water content to become evident. In particular, there are anomalous samples which are either water-rich (up to 1 wt.%) or water-poor (∼0.2 wt.%), with the former being interpreted as forming intrusively within hyaloclastite and the latter representing batches of magma that were volatile-poor prior to eruption. The large data set also provides further insights into the strengths and weaknesses of using volatiles to infer palaeo-ice thicknesses and highlights many of the uncertainties involved. By using examples from Bláhnúkur, the quantitative use of this technique is evaluated. However, the relative pressure conditions which have shed light on Bláhnúkur’s eruption mechanisms and syn-eruptive glacier response show that, despite uncertainties in absolute values, the volatile approach can provide useful insight into the mechanisms of subglacial rhyolitic eruptions, which have never been observed.