Analysis of post-SM4 COS MIRRORB data have revealed that the pre-SM4 throughput values adopted for MIRRORB were too high by a factor 1.7. The main issue was that counts from the secondary peak in MIRRORB images was mistakenly included when calculating sensitivities. A corrected MIRRORB throughput reference file was implemented within SYNPHOT in October 2009, and the Cycle 18 COS Imaging ETC and Imaging Target Acquisition ETC are using the correct throughput values. The impact is that exposure times to reach a particular signal-to-noise ratio with MIRRORB are now longer by a factor of approximately 1.7 as compared to the output of the Cycle 17 version of the COS ETCs. Proposers who intend to recycle calculations performed for a Cycle 17 COS proposal are advised to take this change into account for their Cycle 18 proposal.