Estimated travel speed based on spot speed measurements from detectors has been used as a proxy to actual travel speed. Recently, actual travel speeds become available as vehicles with onboard electronic devices provide actual section travel times. As a result, two different speeds can be estimated for a freeway section at a time interval: estimated travel speed takes the average of the spot speeds collected from each detector within a section at a time interval, and actual travel speed is estimated using the sampled vehicles' actual journey speeds over the section. This paper compares the difference between estimated and actual travel speeds using both kinematic wave theory and actual observations for different traffic states. The theoretical analysis proves that there is a systematic relationship between the estimated and actual travel speeds that form a well-defined loop shape. The existence of well-defined relations is verified using real traffic speed data collected from a Korean expressway and a California freeway. Empirical analysis shows that the real data of estimated versus actual travel speeds exhibit a well-defined linear relationship rather than loop shape. The analysis also shows that the estimated travel speed is generally greater than the actual travel speed due not only to the systematic difference between estimated and actual travel speeds but also to the inherent characteristics of speed data from loop detectors. This finding provides a groundwork to calibrate the widely used estimated travel speed by exploiting the actual travel speeds that were provided by a subset of vehicles in traffic.