Textural characteristics of recently emplaced volcanic materials provide information on the degassing history, volatile content, and future explosive activity of volcanoes. Thermal infrared (TIR) remote sensing has been used to derive the micron-scale roughness (i.e., surface vesicularity) of lavas using a two-component (glass plus blackbody) spectral deconvolution model. We apply and test this approach on TIR data of pyroclastic flow (PF) deposits for the first time. Samples from two PF deposits (January 2005: block-rich and March 2000: ash-rich) were collected at Bezymianny Volcano (Russia) and analyzed using (1) TIR emission spectroscopy, (2) scanning electron microscope (SEM)-derived roughness (profiling), (3) SEM-derived surface vesicularity (imaging), and (4) thin section observations. Results from SEM roughness (0.9-2.8 mm) and SEM vesicularity (18-26%) showed a positive correlation. These were compared to the deconvolution results from the laboratory and spaceborne spectra, as well as to field-derived percentages of the block and ash. The spaceborne results were within 5% of the laboratory results and showed a positive correlation. However, a negative correlation between the SEM and spectral results was observed and was likely due to a combination of factors; an incorrect glass end-member, particle size effects, and subsequent weathering/reworking of the PF deposits. Despite these differences, this work shows that microscopic textural heterogeneities on PF deposits can be detected with TIR remote sensing using a technique similar to that used for lavas, but the results must be carefully interpreted. If applied correctly, it could be an important tool to map recent PF deposits and infer the causative eruption style/mechanism.