Friction and stick/slip phenomena are a major source of disturbance in assembly tasks, especially under low velocity or frequent start/stop conditions. These friction effects are highly nonlinear, dependent on environmental parameters, and the resulting forces are extremely difficult to model accurately under all conditions. This paper discusses frictional contacts in the context of robotic assembly based on the underlying physics of the phenomenon. A technique employing a high frequency dither is presented which reduces the effect of friction between parts in relative motion. The parameters of this dither are tuned on-line to account for changes in the operating conditions. This tuning algorithm adjusts the dither parameters based on force feedback measurements and a model for the desired forces. Analysis shows that a set of optimal parameters exists. Experimentation demonstrates the algorithm's success, even in the presence of uncertain or varying system model parameters. The discussion includes consideration of previous friction studies and combination with force-guided control.