Galactic black holes (GBH) radiate in one of several spectral states,
switching from one to another for reasons that are as yet not understood.
The long standing questions are the geometry of the X/ 
-ray emitting
region (``corona'') and the radiative mechanisms generating the spectrum.
There is a consensus that thermal Comptonization is responsible for the
hard state spectra of these sources that cut off at 
100 keV. The
origin of the soft state spectra extending up to 1 MeV is a matter
of debate. A hybrid, thermal/non-thermal model was developed where such
spectra are produced by Comptonization of soft photons from the cool
accretion disc by basically non-thermal electrons in the corona.
An important feature in the GBH spectra is the so called Compton reflection component that arises due to reflection/reprocessing of coronal emission by the accretion disc. The observed small amplitude of this component (the lack of reflector) was interpreted as a signature of the transformation of the thin accretion disc in the vicinity of the black hole into a hot geometrically thick flow emitting hard X-rays. However, the physical reasons for such a transition remain unclear.
An alternative model was then suggested: the lack of reflection is due to
mildly relativistic bulk motion ( 
) of the coronal plasma away
from the reflector. Bulk motion causes aberration reducing the X-ray
emission towards the disc, which leads to a small reflection. The model
predicts a correlation between the spectral slope and the amount of
reflection that successfully reproduces the recently observed correlation.
These findings favour the picture where the corona consists of compact
magnetic flares dominated by 
pairs which are ejected away from the
reflector by the pressure of the reflected radiation.
The picture of energy release in compact magnetic flares helps to resolve a long standing mystery of large time lags between coronal emission at different X-ray energies. These lags reach a few tenths of a second, exceeding by orders of magnitude the light-crossing time of the emitting region. It was proposed that the time lags are related to the time scales of the spectral evolution of individual magnetic flares which may be comparable to the Keplerian time scales at 3-100 gravitational radii from the central black hole. The overall temporal variability was then modelled using a flare avalanche model in which each flare has a certain probability to trigger a neighbouring flare, thus occasionally producing long avalanches. The proposed model describes well the time-averaged spectra as well as temporal characteristics such as the power-density spectrum, the time/phase lags, and the coherence function of Cygnus X-1.