Participants of a large-scale, real-life peak avoidance experiment have been asked to provide estimates of their average in-vehicle travel time for their morning commute. Comparing these reported travel times to the corresponding actual travel times, we find that travel times are overstated by a factor of 1.5 on average. We show that driver- and link-specific characteristics partially explain the overstating. Using stated and revealed preference data, we investigate whether the driver- specific reporting errors are consistent with the drivers' scheduling behavior in reality as well as in hypothetical choice experiments. For neither case, we find robust evidence that drivers behave as if they misperceived travel times to a similar extent as they misreported them, implying that reported travel times do neither represent actual nor perceived travel times truthfully. The results presented in this paper are thus a strong caveat against the uncritical use of reported travel time data in transport research and policy.