Observations of the energetic ion composition in the magnetosphere are reviewed with emphasis on the recent measurements by means of GEOS 1 and 2, ISEE 1 and 2, PROGNOZ 7, and SCATHA. The observations are compared with the predictions of the open magnetosphere model. One of the major conclusions is that there are processes in the magnetosphere which play a much larger part than the model, as hitherto presented, predicts. Direct ejection of ionospheric ions, in combination with acceleration, along closed as well as open field lines may even be the dominant source process for the ring current/inner plasma sheet in magnetic storms. In very disturbed conditions this ejection mechanism must work over most of the hemisphere poleward of, say, 50°. Circulation of the ionospheric ions through the tail of the magnetosphere is not likely to be of primary importance for the energization of these ions in very disturbed conditions. Other conclusions about the ring current/plasma sheet plasma are as follows: (1) The polar wind may be the dominant source of ring current ions in the keV range near the geostationary orbit in quiet conditions. (2) The detached plasma blobs from the plasmasphere do not generally appear to be an important source of hot ions in the inner magnetosphere. (3) In the inner ring current region (L < 4), charge exchange loss processes give rise to the dominance in this region of ions of ionospheric origin in quiet conditions and during the recovery phase of magnetic storms. (4) Acceleration both along and perpendicular to field lines affects the ejected ionospheric ions. (5) Mass‐dependent acceleration processes appear to be of importance; ion beams which may have traveled adiabatically from below a few thousand kilometers altitude to 5 RE along the field lines have been seen, but mostly, the beams are wide at great heights. (6) Energized electrons are sometimes found flowing along the same field tubes as the energized ions. (7) There appear to exist more or less simultaneously quite different source mechanisms in different parts of the magnetosphere. (8) There is a solar cycle variation in the ion composition. Conclusions from the recent investigations of the composition of the magnetospheric boundary layers by means of the PROGNOZ 7 satellite are as follows: (1) The dayside low‐latitude boundary layer/entry layer contains a structure of unmixed regions with plasma of either hot ‘magnetospherelike’ characteristics or less hot and more ‘magnetosheathlike’ properties within the boundary layer, but the magnetosheathlike plasma also contains ions of ionospheric origin. (2) The plasma mantle contains large, although generally not dominant, amounts of ions of ionospheric origin in disturbed conditions but frequently no such ions in quiet conditions. (3) The mantle frequently contains ‘inclusions’ of high‐density (β ≥ 1) magnetosheathlike plasma with sometimes strongly deviating magnetic field and even flow and with Birkeland currents at their edges. (4) The flank mantle frequently contains stagnant plasma with appreciable amounts of ionospheric ions of hard energy spectra and also strongly heated electrons. (5) Mostly, the mantle observations appear to be consistent with both acceleration processes which provide about equal velocity to the ions, irrespective of mass (thought to be associated with reconnection), and processes which give equal E/q to the ions, irrespective of mass (electrostatic potential differences), but they do not generally demonstrate which of these two kinds of processes, or any other, has accelerated the ions; cases have, however, been found in which the various ion species had equal E/q. (6) At least in disturbed conditions the plasma mantle is not the dominant source of the hot plasma in the ring current/plasma sheet. (7) The nightside magnetopause appears to be a more solid boundary for the heavy ions of ionospheric origin than is expected from the open magnetosphere model. (8) In other respects the observations in the deep nightside mantle (far from the flanks) are in general in reasonable agreement with the open magnetosphere model, if this is amended with ‘additional processes’ which frequently increase the ionospheric ion content in the mantle.