In this dissertation, the results of experimental and theoretical studies exploring friction and adhesion at the nanoscale are presented. Using a customized in situ transmission electron microscopy nanoindentation methodology, it is observed that cohesion of silicon and adhesion of silicon and diamond are strongly modied by the sliding speed and the normal stress applied during sliding. This indicates that shear stress modulates the reactivity of the surfaces. This is the rst time that tunable adhesion of hard contacts has been demonstrated in situ. If sliding experiments are performed in ultra-high vacuum and the interfacial shear stress is low enough to avoid surface modication, the Multibond model of friction predicts that adhesion will decrease with increasing sliding speed in experiments with simultaneous sliding and retraction. Results from sliding of nanoscale silica asperities against highly-oriented pyrolytic graphite (HOPG) and hydrogen-doped tetrahedral amorphous carbon (a-C:H) surfaces are consistent with this model. This contrasts with the directly-proportional adhesion-speed behavior observed in the in situ transmission electron microscopy experiments of silicon and diamond. When the number of available bonding sites increases with stress and speed, adhesion will increase. This is the case for the silicon-silicon and silicon-diamond work. However, if the number of available sites is constant, sliding faster will further reduce adhesion. This is the case of the work of silica sliding against HOPG and a-C:H. Existing popular reduced order models for friction, the Prandtl-Tomlinson with temperature model and the Multibond model, are frequently used to explain the observed nanoscale phenomena of friction increasing logarithmically with sliding speed. However, both models contain overgeneralizing or unphysical assumptions. A new model, the modied Multibond model, was developed and is consistent with experimental results. This dissertation provides strong evidence that damping is a critical parameter and that the Fokker-Planck equation is more suitable to describe friction-speed behavior than the Prandtl-Tomlinson with Temperature and Multibond models. The modied Multibond model also predicts the decrease of adhesion with increasing speed observed experimentally in the silica-HOPG and silica a-C:H experiments.