Although a broad range of studies reveal that occupants move their blinds far less frequently, current manual blind control algorithms are too active and often overestimate the way people adjust manual blinds. Accordingly, a longitudinal human factors field study was conducted based on observations including 92,480 blinds position recordings of 7,080 window groups in 6 mid-rise office buildings located in Boise, Idaho to provide more accurate predictions of actual user behavior and develop a revised manual blind algorithm - called Blindswitch 2018 (updated version of Blindswitch-2017). Results revealed that roughly 94 % of the monitored blinds were not adjusted during a given average day, and 50% of blinds didn't have any movement during the entire period of study. Moreover, the findings showed that only 10% of occupants are active in adjusting blinds, while large portion of occupants behave as “non-users” (50%) or “passive users” (40%). In comparison to current “manual” blind use algorithms that assume all occupants as active users, the proposed algorithm would be a significant step forward to accurately simulate actual user behavior in actual buildings. Blindswitch-2018 algorithm is considered to be an improved basis for predicting the state of manual blinds in office buildings and can be implemented as a more realistic baseline to estimate overall performance of manual blinds in energy, daylighting, and electric lighting simulations.