Mass loss is a key process in the evolution of massive stars, and must be
understood quantitatively to be successfully included in broader astrophysical
applications. In this review, we discuss various aspects of radiation driven
mass loss, both from the theoretical and the observational side. We focus on
winds from OB-stars, with some excursions to the Luminous Blue Variables, Wolf-
Rayet stars, A-supergiants and Central Stars of Planetary Nebulae. After reca-
pitulating the 1-D, stationary standard model of line-driven wind, extensions
accounting for rotation and magnetic fields are discussed. The relevance of the
so-called bi-stability jump is outlined. We summarize diagnostical methods to
infer wind properties from observations, and compare the results with theore-
tical predictions, featuring the massloss-metallicity dependence. Subsequently,
we concentrate on two urgent problems which challenge our present understanding
of radiation driven winds: weak winds and wind- clumping. We discuss problems
of measuring mass-loss rates from weak winds and the potential of NIR-
spectroscopy. Wind-clumping has severe implications for the interpretation of
observational diagnostics, as derived mass-loss rates can be overestimated by
factors of 2 to 10 if clumping is ignored, and we describe ongoing attempts to
allow for more uniform results. We point out that independent arguments from
stellar evolution favor a moderate reduction of present- day mass-loss rates.
We also consider larger scale wind structure, interpreted in terms of
co-rotating interacting regions, and complete this review with a discussion of
recent progress on the X-ray line emission from massive stars, highlighting as
to how far the analysis of such X-ray line emission can give further clues
regarding an adequate description of wind clumping. (Abridged abstract)Comment: Astronomy and Astrophysics Review (accepted