Minimizing the time needed for aircraft to taxi from gates to runways is an important part of efforts to manage the national airways system. For that reason, the Federal Aviation Administration (FAA) currently collects detailed information on all ground traffic in order to help estimate delays in taxiing out to runways. The agency bases its definition of taxi-out delay on the difference between actual taxi-out times and a constructed value referred to as “unimpeded taxi-out time.” Currently, unimpeded times are estimated applying a statistical relationship between historical taxi-out times and the levels of outgoing and incoming traffic under the further assumption of “ideal” ground conditions: that is, (1) the weather is optimal, and (2) just one flight is leaving and no flights are arriving. In a completely different way, using a novel approach that groups flights by airport, airline, season, day, and hour in order to control for variations in taxiing conditions, the paper would use FAA information on the observed minimum of actual taxi-out times to identify a new “time reference” estimate of the unimpeded taxi-out times. The paper describes criteria for identifying and discarding data errors that hamper the current FAA data. It presents an example of the estimation for one time reference grouping. Importantly, the paper also describes how to provide a measure of the uncertainty in that deterministic estimate of the unimpeded time by hypothesizing and testing a probability distribution for those times and presenting examples of how those probability distributions can be used to make inferences.