We used a combination of Goldstein (a classic as another poster pointed out) and Marion & Thornton's Classical Dynamics of Particles and Systems (which was good but not great).