Current day routing systems use predicted travel times to calculate optimal routes. Some routing algorithms even use predicted traffic lights states to optimize these routes and to improve the expected travel time. But all these methods optimize the expected travel time and do not take reliability into account. Reliability is a major service indicator and with non-flexible departure or arrival constraints, the value of reliability can increase up to three times the value of time (Markovich, Concas, & Kolpakov, 2009). To use a measure of reliability, not the expected travel time, but the travel time distribution should be known. Since reliability is not uniquely defined, an algorithm is designed with an individual cost function, that gives users the freedom to use their own definition of reliability. Since knowing the traffic light states at intersections could improve the travel time and reliability, this should be used in the algorithm as well. Therefore the main goal of this research is: Designing an urban routing algorithm, that incorporates predicted traffic light states, where the routing is based on the probability distribution of the travel time and an individual cost function that expresses the tradeoff between travel time and reliability. For this research only urban areas with traffic lights are considered, other intersections in this network are not taken into account. The traffic lights are assumed to be traffic lights with a changing cycle time. If this algorithmworks for these traffic lights it will also work for fixed time traffic lights and most other type of traffic lights, since these have less degrees of freedom.