The Edinburgh-Cape Blue Object Survey is an ongoing project to identify and analyse a large sample of hot stars selected initially on the basis of photographic colours (down to a magnitude limit (B~18.0) over the entire high-Galactic-latitude southern sky, and then studied with broad-band UBV photometry and medium-resolution spectroscopy. Due to unavoidable errors in the initial candidate selection, stars that are likely metal-deficient dwarfs and giants of the halo and thick-disc populations are inadvertently included, yet are of interest in their own right. In this paper we discuss a total of 206 candidate metal-deficient dwarfs, subgiants, giants, and horizontal-branch stars with photoelectric colours redder than (B-V)_0_=0.3, and with available spectroscopy. Radial velocities, accurate to ~10-15km/s, are presented for all of these stars. Spectroscopic metallicity estimates for these stars are obtained using a recently recalibrated relation between Ca II K-line strength and (B-V)_0_ colour. The identification of metal-poor stars from this colour-selection technique is remarkably efficient, and competitive with previous survey methods. An additional sample of 186 EC stars with photoelectric colours in the range -0.4<=(B-V)_0_<0.3, photoelectric colours in the range composed primarily of field horizontal-branch stars and other, higher gravity, A- and B-type stars, is also analysed. Estimates of the physical parameters T_eff_, log g and [Fe/H] are obtained for cooler members of this subsample, and a number of candidate RR Lyrae variables are identified.