This paper reports user requirements for frequency standards for optical communications and reviews physical possibilities for measurement standards to support those needs. A survey of the industrial and regulatory requirements for frequency standards for optical communications was made between October 1991 and January 1992. This was centred on the UK and Europe, but also took in responses from the USA and Japan. Over 70 representatives from 49 organisations were contacted, and a response exceeding 50% was achieved. The main requirement found was a need for two frequency standards per band in each of the 1.5 jim and 1.3 jim bands, with those in the 1.5 jim band needed first, and in the next 3-5 years. An curacy of order 1 part in 10' was desired for laboratory use, with an order of magnitude less for a transfer standard, and an order less again in the field.The physical possibilities for frequency standards in these bands are reviewed, addressing the use of atomic and molecular resonances. In these bands transitions in atoms are from excited states, and in molecules they are either overtone or combinationband transitions. This makes the transitions difficult to saturate, to obtain sub-Doppler feature suitable for 1 part in 1O frequency standards. A summary table is presented of the stronger relevant spectra from the literature and from recent high-resolution Fourier transform spectroscopy at the University of Strathclyde. A further table lists frequency standards already measured mainly by heterodyne techniques, and which are mainly in the 1.5 i.un band. The potential adequacy of the standards to meet the requirements, and a hierarchy for band sub-division in frequency are discussed. 106