Friction stir welding (FSW) is a solid-state joining process for many materials including aluminum alloys, steel, titanium, and plastics. However, attendant process variability such as uneven workpiece flatness and workpiece temperature changes is known to lead to joints with inconsistent material properties. The weld quality can be improved by monitoring and controlling the spindle power. This work presents a multi-level adaptive fuzzy control approach for maintaining constant FSW power. Implementation of the control scheme is conducted on a vertical milling machine configured for FSW using an open architecture controller. Experimental results show that the multi-level fuzzy technique dramatically reduces power variations in the presence of significant process disturbances.