Many cities are dealing with traffic problems. One of emerging of these problems is that travel times are longer than estimated values. Variability in travel times leads to disruptions in supply chain and delayed passengers, amongst others. Variability of travel time is a phenomenon dependent on time and space. Determining this variability is important for transportation planning in traffic-intensive times and regions. Such a quest requires big geospatial data. The taxi dataset of New York City has been openly distributed since 2009, which describes each taxi trip with attributes such as the pickup and dropoff location and time, total cost of the trip and number of passengers. The aim of this article is to determine the travel time variability and distributions between John F. Kennedy and LaGuardia airports using the openly available taxi dataset involving approximately 140 million trips that occurred in 2015. The journeys, which take place every day of the week and between 07:00 and 19:00, are analyzed at 15 minute intervals. Log-Normal, Log-Normal (3P), Log-Logistics, Log-Logistics (3P), Weibull, Gamma and Burr distributions were examined. 95% of travel time, buffer time index and planned time index were used as the travel time reliability measurements. According to the results obtained, travel times are consistent with Log-Logistics (3P) distribution. This distribution is not the dominant distribution on Tuesdays and Wednesdays and it has been found that there are times when travel time reliability is low.