Precision-controlled mechanisms are commonly used in machining applications where the feed rates are low, direction reversals are necessary to obtain the desired work piece profile and precision requirements are in the order of micrometers or sub-micrometers. Under these conditions, friction and gear backlash effects contribute significantly to the dynamics of the system. This study investigated the use of neural network models to compensate for these effects. The approach was to switch between two controllers: (i) a proportional-plus-derivative controller together with a feedforward friction compensator when the transmission gears were engaged; and (ii) a feedforward backlash controller when the transmission gears were disengaged. The control scheme was experimentally verified using a retrofitted geared head engine lathe. The results obtained were compared to those obtained with a compensation method found in the literature.