There are no satisfactory MEMS vacuum pumps; particularly unavailable are vacuum pumps that can handle the flow required for MEMS scale, continuous sampling instruments. The two obvious paths to creating such pumps, adapting current technology to MEMS scales or inventing new MEMS friendly technology, are discussed. The first path has been tried to some extent and not been successful, the second has been studied and a few possibilities are under investigation. A generic scaling study of expected trends in vacuum pumping performance as size is reduced to the MEMS scale is presented. It indicates that in practically all cases present vacuum pumps scaled to MEMS dimensions are not very attractive. New technologies that may offer more attractive possibilities are discussed. The degree of attraction is measured in terms particularly applicable to MEMS devices; the energy required per unit of upflow in the pump and the pump volume per unit of upflow. Both of these need to be sufficiently small to permit self consistency in energy use and size in order to allow local integration of the pumps with the MEMS devices that require vacuum pumping. For instance the full potential of instruments such as a MEMS sampling mass spectrometer can only be achieved if the pumps have power or space requirements equal to or preferably significantly less than the instrument itself. It is concluded that with new pumping technologies that have been identified, it may be possible to provide satisfactory MEMS vacuum pumping performance. However, this will only come to pass if a determined research and accompanying development program is created.