Collection of faecal samples for microbiome analysis in acutely sick patients is logistically difficult, particularly if immediate freezing is required (i.e. fresh-frozen, or FF sampling). Previous studies in healthy/non-hospitalized volunteers have shown that chemical stabilization (i.e. stabilized-frozen, or SF sampling) allows room-temperature storage with comparable results to FF samples. To test this in a hospital setting we compared FF and SF approaches across 17 patients undergoing haematopoietic stem cell transplantation (HSCT) using both 16S rRNA gene and shotgun metagenomic sequencing. A paired (same stool specimen) comparison of FF and SF samples was made, with an overall comparable level in relative taxonomic abundances between the two sampling techniques. Though shotgun metagenomic sequencing found significant differences for certain bacterial genera (P < 0.001), these were considered minor methodological effects. Within-sample diversity of either method was not significantly different (Shannon diversity P16SrRNA = 0.68 and Pshotgun = 0.89) and we could not reject the null hypothesis that between-sample variation in FF and SF were equivalent (P16SrRNA = 0.98 and Pshotgun = 1.0). This indicates that SF samples can be used to reliably study the microbiome in acutely sick patient populations, thus creating and enabling further outcomes-based metagenomic studies on similarly valuable cohorts.