We present the results of a redshift survey of 138 candidate compact groups from the DPOSS II catalogue, which extends the available redshift range of spectroscopically confirmed compact groups of galaxies to redshift z ∼ 0.2. In this survey, we aim to confirm group membership via spectroscopic redshift information, to measure the characteristic properties of the confirmed groups, namely their mass, radius, luminosity, velocity dispersion, and crossing time, and to compare them with those of nearby compact groups. Using information available from the literature, we also studied the surrounding group environment and searched for additional, previously unknown, group members, or larger scale structures to whom the group might be associated. Among the 138 observed groups, 96 had 3 or more concordant galaxies, i.e. a 70% success rate. Of these 96, 62 are isolated on the sky, while the remaining 34 are close on the sky to a larger scale structure. The groups which were not spectroscopically confirmed as such turned out to be couples of pairs or chance projections of galaxies on the sky. The median redshift of all the confirmed groups is z ∼ 0.12, which should be compared with the median redshift of 0.03 for the local sample of Hickson compact groups. The average group radius is 50 Kpc, and the median radial velocity dispersion is 273 km s −1 , while typical crossing times range from 0.002 H −1 0 to 0.135 H −1 0 with a median value of 0.02 H −1 0 , which are quite similar to the values measured for the Hickson compact group sample. The average mass-to-light ratio of the whole sample, M/L B , is 192, which is significantly higher than the value measured for Hickson's compact groups, while the median mass, measured using the virial theorem, is M = 1.67 × 10 13 M . When we select only the groups that are isolated on the sky, the values quoted become lower and closely resemble the average values measured for Hickson's compact groups. We also find that the characteristics of the groups depend on their environment. We conclude that we observe a population of compact groups that are very similar to those observed at zero redshift. Furthermore, a careful selection of the environment surrounding the compact groups is necessary to detect truly isolated compact structures.