We present a study of the deuterium abundance along the extended sight line toward HD 90087 with the Far Ultraviolet Spectroscopic Explorer (FUSE). HD 90087 is a O9.5 III star located in the Galactic disk at a distance of ~2.7 kpc away from the Sun. Both in terms of distance and column densities, HD 90087 has the longest and densest sight line observed in the Galactic disk for which a deuterium abundance has been measured from ultraviolet absorption lines so far. Because many interstellar clouds are probed along this sight line, possible variations in the properties of individual clouds should be averaged out. This would yield a deuterium abundance that is characteristic of the interstellar medium on scales larger than the Local Bubble. The FUSE spectra of HD 90087 show numerous blended interstellar and stellar features. We have measured interstellar column densities of neutral atoms, ions, and molecules by simultaneously fitting the interstellar absorption lines detected in the different FUSE channels. As far as possible, saturated lines were excluded from the fits in order to minimize possible systematic errors. IUE (International Ultraviolet Explorer) archival data are also used to measure neutral hydrogen. We report D/O = (1.7 ± 0.7) × 10-2 and D/H = (9.8 ± 3.8) × 10-6 (2 σ). Our new results confirm that the gas-phase deuterium abundance in the distant interstellar medium is significantly lower than the one measured within the Local Bubble. We supplement our study with a revision of the oxygen abundance toward Feige 110, a moderately distant (~200 pc) sdOB star, located ~150 pc below the Galactic plane. Excluding saturated lines from the fits of the FUSE spectra is critical; this led us to derive an O I column density about 2 times larger than the one previously reported for Feige 110. The corresponding updated D/O ratio on this sight line is D/O = (2.6 ± 1.0) × 10-2 (2 σ), which is lower than the one measured within the Local Bubble. The data set available now outside the Local Bubble, which is based primarily on FUSE measurements, shows a contrast between the constancy of D/O and the variability of D/H. As oxygen is considered to be a good proxy for hydrogen within the interstellar medium, this discrepancy is puzzling.