All layered manufacturing processes operate by trading off between part accuracy and fabrication time. In particular, this is manifested in the strategy for generating the geometry of each slice of the 3D model: if the slice thickness is large, the fewer slices can build the part, taking up less time, but the surface roughness (and thus part accuracy) is poor. Since the 1990’s, several slicing approaches have been proposed to reduce the printing time while improving the accuracy. However, there is no optimal strategy known to date. On the one hand, the high computational complexity of some algorithms renders them impractical. On the other hand, adaptive slicing approaches have not been successfully implemented on any commercial machine due to problems in material and process control. In the thesis, a novel adaptive slicing algorithm based on slice contour reconstruction is proposed and implemented. This approach can be adapted for standard additive manufacturing machines because it does not rely upon adaptive layer thickness. At the same time, a simple geometric engine can allow our approach to be transparently adapted to any existing machine by re-creating an updated 3D product model using the modified layers. The contour reconstruction algorithm based on Medial axis computation leads to 30%-50% reduction on both volumetric deviation and maximum horizontal distance deviation between the original triangulated part surface and the deposited part surface. Alternatively, for a given user-specified surface finish, our approach allows for larger layer thickness and consequently produces time saving of 20-30% over the traditional slicing strategy.