In the Team Orienteering Problem (TOP), a team of vehicles attempts to collect rewards at a given number of stops within a specified time frame. Once a vehicle visits a stop and collects its reward, no other vehicles may collect the reward. Typically, a team cannot visit all stops and therefore has to identify the “best” set of stops to visit in order to maximizes total rewards. We propose a large neighborhood search method with three improvement algorithms: a local search improvement, a shift and insertion improvement and replacement improvement. Our proposed approach can find the best known solutions for 386 of the 387 benchmark instances; for the one instance which our solution is not optimal it is only varies by one from the current best solution. Other methods are comparable in terms of solution quality, however, our approach has significant benefits in terms of computational time.