Up to the present time flame media have been used almost exclusively as atom reservoirs for the purposes of atomic absorption and atomic fluorescence spectroscopy. The merits and demerits of flames in various forms will be discussed against the differing requirements of each technique. Particular attention will be paid to the use of the separated flame in each mode of analysis. Flame emission spectroscopy will also be discussed briefly in relation to this topic.Several alternatives to the flame have been proposed notably the r.f. plasma torch, Langmuir-torch, the sputtering (hcllow cathode lamp) chamber, the Vidale (King) furnace, L'vov furnace and Massmann tube. These devices will be briefly reviewed and the F AR (filament atom reservoir) unit devised at the Imperial College will be discussed more fully The device fits most commercial equipment in place of the normal flame/nebulizer unit and allows very high sensitivity, e.g. 10-10 g Mg in absorption, 10-15 g Mg in fluorescence etc. It is self-purging and extremely simple to use in practice. A proposal is made relating to terminology for the sensitivity ofnon-nebulizing methods ofatomic absorption and atomic fluorescence analysis.The use of microwave-excited electrodeless discharge tubes as spectral line sources for individual elements in atomic absorption and atomic fluorescence spectroscopy will be discussed and the development of multi-element discharge tubes for sequential atomic fluorescence using an integrating flame photometer with automatic flame background c::~rrection.