SCIENTIFIC ABSTRACT (A) We propose to derive the C, N and O abundances in the outer winds of WR stars and O and B-type stars by measuring, in LWS grating mode, the fluxes in the fine structure lines of [C II], [N II], [N III] and [O III]. Such abundances cannot easily be derived from optical or ultraviolet spectra. The WC9 stars and some of the extreme supergiants show strong circumstellar dust emission, whose far-IR spectral characteristics will be simultaneously measured. Three of the extreme B supergiants have been found to show 2.3um CO emission by McGregor et al. (1988), so CO rotational line emission may be detectable in their LWS spectra. B) The shells of old novae are interesting in several respects, most notably because they are often highly enriched in heavy elements of the oxygen group and beyond. The most important objective of ISO for novae is determining the elemental abundances in the ejecta, from observations of fine structure lines. IR observations are essential because, due to the enrichment, the gas easily cools to very low temperatures, so that the optical and UV forbidden lines eventually become very weak. (C) Novae occurring during the ISO mission are likely to be targets of opportunity. However, young novae that have occurred in the year or so before launch and during the ISO mission could also be profitably observed, particularly in order to follow the rise in excitation of the fine structure lines as the central remnant evolves to higher temperatures. Such observations would also help to determine whether the dramatic fall in [O III] 5007A emission in some evolving nova shells is due to strong evolution and fading of the ionization source, or is instead due to the temperature drop caused by the increased efficiency of IR fine structure line cooling as the density of the expanding shell drops below the critical density for collisional de-excitation of the upper levels of the various lines. In addition, ISO observations with both LWS and SWS can for the first time provide a complete study of the recently discovered infrared coronal emission line phase of young classical nova outbursts, resulting in new constraints on heavy element abundances in both their ejecta and the cores of their precursor stars. A LaTex file containing fuller details of this proposal is available from OBSERVATION SUMMARY For either a Spring or Autumn launch, 8.95 hours (535.1 minutes) of LWS Guaranteed (spacecraft) time are allocated to this programme. For a Spring launch (Orion-hole observable), 474.5 minutes will be used for full LWS01 43-196.7um grating scans of 17 early-type stars, 5 old novae and 2 young nova, using 4 sample points per resolution element and an integration time of 2 seconds per sample point, along with 60.6 minutes for full SWS01 2.43-45um low-resolution (R/8) grating scans of the 2 young novae. For an Autumn launch (Orion-hole observable) 474.5 minutes will be used for full LWS01 43-196.7um grating scans of 16 early-type stars, 6 old novae and 2 new nova, along with 60.6 minutes for full SWS01 2.43-45um R/8 grating scans of the 2 new novae. The Priority 1 LWS Guaranteed spacecraft time allocated to this programme is 3.55 hours for the case of either a Spring or an Autumn launch.