Filter-based BC measurement techniques such as the Continuous Soot Monitoring System (COSMOS) are particularly well suited to long-term observations of black carbon (BC) due to their relative robustness and reliability. However, caution is required when determining the threshold transmittance, Tr-thresh (roughly proportional to the time interval between filter changes), in order to ensure that acceptable measurement accuracy is maintained throughout the sampling period. We present a new, empirically derived transmittance-dependent correction factor used to interpret the response characteristics of filter-based aerosol absorption measurements performed by COSMOS. Simultaneous measurements of ambient BC aerosol mass (M-BC) were conducted in Tokyo, Japan, using two identical COSMOS instruments operated with different threshold transmittance, Tr-thresh, values, of 0.95 and 0.6. The derived values for M-BC were consistently underestimated by the COSMOS operating at lower Tr-thresh, as a function of decreasing filter transmittance. The 1-hour averaged values of M-BC were underestimated by around 10%, incorporating measurements across the entire range of filter transmittance (1-0.6), with a maximum underestimation at around 17% immediately preceding filter advancement (i.e. Tr=similar to 0.6). An empirical correction factor was derived from these ambient measurements, and was applied to M-BC as a function of filter transmittance, resolving the instruments to within 2%. Further to the transmittance-based correction, the operational performance of COSMOS was tested for two types of quartz fibre filter (PALLFLEX and HEPA). Agreement in derived values of M-BC for two COSMOS using the same type of filter was around 2%; however, a comparison of the PALLFLEX and HEPA filters demonstrated a systematic overestimation of M-BC derived when using HEPA filters, of around 6-8%. A sensitivity study of a radiative transfer model indicated that this enhanced absorption was primarily a result of the increased thickness of the HEPA filter. (C) 2014 Elsevier Ltd. All rights reserved.