Among all forms of traveler information, travel time is regarded as the most critical because it can empower drivers to make more informed decisions and potentially improve the performance of the entire system. Numerous studies have been conducted for predicting travel time based on measured data, mostly from loop detectors. This paper focuses on evaluating the performance of a set of “benchmark” methods designed to utilize loop data to provide route travel times for display on Changeable Message Signs (CMS). The paper reveals that (1) although the “ground-truth” travel time (e.g., those obtained from probe vehicles) at a given time instant is often considered as a single value, it may be more appropriately represented as an interval to reflect the travel time variation due to different driving aggressiveness; (2) lane-by-lane loop detector data may be better utilized to improve travel time estimation, and (3) the performances of loop travel times vary dramatically during different time-of-day (TOD), and (4) as detector spacing increases, the accuracy of travel time estimation decreases, while its variation increases significantly.