The present generation of large scale dE/dx detectors was conceived in the mid-seventies. In fixed target physics applications (EPI, ISIS) the achieved particle identification efficiency allows for ¢r/K separation in the relativistic rise region. In the family of compact colliding beam detectors (e.g. AFS, ARGUS, CLEO, JADE, TPC, UAI), only the LBL TPC performance was so far sufficient for operation beyond the minimum of ionization. New central detectors foreseen for experiments in near future (LEP) should be more performing and will require resolutions which are at the limits imposed by systematic higher order effects. Influence of various design parameters on the optimization of the dE/dx sampling performance is discussed, supported by recent result from measurements of full-scale detector segments and including tests using ionized tracks generated by pulsed UV lasers. Merits of longitudinal drift and fine sampling are compared to results obtained by classical charge integration,