Measurements are reported of the main factors, namely the AC voltage frequency and magnitude, that were observed to influence the number of cells destroyed during dielectrophoresis (DEP) experiments on Jurkat T cells and HL60 leukemia cells. Microelectrodes of interdigitated and quadrupolar geometries were used. A field-frequency window has been identified that should be either avoided or utilised, depending on whether or not cell damage is to be minimised or is a desired objective. The width and location of this frequency window depends on the cell type, as defined by cell size, morphology and dielectric properties, and is bounded by two characteristic frequencies. These frequencies are the DEP cross-over frequency, where a cell makes the transition from negative to positive DEP, and a frequency determined by the time constant that controls the frequency dependence of the field induced across the cell membrane. When operating in this frequency window, and for the microelectrode designs used in this work, cell destruction can be minimised by ensuring that cells are not directed by positive DEP to electrode edges where fields exceeding 30-40 kV/m are generated. Alternatively, this field-frequency window can be exploited to selectively destroy specific cell types in a cell mixture.