A diverse and compelling array of experiments has led to the emergence of
a `concordance' cosmological model, the Lambda CDM paradigm.  
Sophisticated models of galaxy formation in this framework, which include
gas cooling, star formation and stellar feedback, have made reasonably
robust predictions about the evolution of the most massive galaxies
(galaxies with high stellar mass): massive galaxies i) should be
reasonably common, even at relatively early times; ii) should have a
number density which rises with time to the present day; and iii) should
still host ongoing star formation.

In this talk, I present an empirical discussion of the evolution of the 
massive galaxy population over the last 8-10 billion years.  In agreement 
with the models, massive galaxies exist even at early times and their 
number density increases constantly to the present day.  In striking 
disagreement with the models, most massive galaxies do not form stars.  
In the last eight to ten billion years, a population of primarily massive 
galaxies which do not form stars has emerged -- the so-called red sequence 
-- whose number has increased to the point that at the present day they 
hold more than half of all stars ever formed.  I present empirical 
constraints on the rate at which galaxies turn off as a function of cosmic 
time.  Finally, I discuss the possible role of galaxy merging --- a 
relatively common occurrence between massive galaxies --- as a trigger for 
the quenching of star formation in massive galaxies.