There is a general consensus regarding the ages of the stellar populations
in early-type cluster galaxies, i.e. in ellipticals and S0s as
morphologically-classified from HST images. The slope, zero-point and
scatter of their color-magnitude (CM) relation at various redshifts
indicate passive evolution of stellar populations that formed at
z >2-3, and the CM slope is found to be mostly driven by a
mass-metallicity relation
([19,
20,
21,
22,
16], see also
[23]).
Moreover, well defined CM red sequences in higher redshift
clusters (z 
1) are
now being observed (e.g.
[24,
25,
26,
27]),
and there are hints for a flatter CM sequence at high-z
[29,
28]
that still needs to be comprehended. The old stellar ages in
early-type galaxies are confirmed also by their spectroscopic features
(see below, e.g.
[30,
31])
and by fundamental-plane, mass-to-light ratio and
Mg-
studies
(e.g. [32,
33,
34,
35,
36,
37,
38]),
though this latter type of works have been
necessarily limited to relatively small numbers of the brightest galaxies.
This homogeneity of histories of early-type galaxies in clusters needs to be contrasted with a number of other results and considerations:
a) The population of early-type galaxies observed in distant clusters does not necessarily comprises all the early-type galaxies existing at z = 0. Morphological transformations might occur in clusters (see below) and change later-type starforming galaxies into some of the early-type galaxies present in clusters today. Due to this "progenitor bias", in distant clusters we would be observing only those galaxies that were already assembled as early-type and stopped forming stars at high redshift [32, 20, 39].
b) The blue galaxies observed in distant clusters, responsible for the Butcher-Oemler effect, must largely have "disappeared" (= become red) by z = 0. [4] (see also [2]) have shown that the color-magnitude diagram observed at intermediate z (rich of blue galaxies) can be reconciled with the CM diagram of the Coma cluster (with very few luminous blue galaxies) if star formation is halted in the blue galaxies at intermediate redshifts.
c) Related to the previous point, some works at z ~ 1 argue for a truncation of the CM sequence [40], as only bright galaxies are found to be in place on the sequence and there is a deficit of fainter red galaxies. This is possibly related to the fact that when star formation is halted in blue Butcher-Oemler galaxies they evolve becoming redder and fainter, going to populate the red sequence at magnitudes fainter than the top brightest 1 or 2 magnitudes. Furthermore, a UV-excess has been found in the galaxies on the red sequence as compared to passive evolution models [18, 40].
d) On the color-magnitude sequence, there are also passive spirals at all redshifts ([31] at z = 0.5, [41] at z = 0.3, [42] in Coma, [43] at z = 0.25) and some mergers at z = 0.8 [44].
All these findings suggest that the color-magnitude sequence of clusters today (except the very brightest end of it) must be composed of a variegated population of galaxies that had a variety of star formation histories. Though probably the mass-metallicity relation remains a main driver of the CM relation, intricate age and metallicity effects must be at work, as also shown by the age-metallicity anticorrelation in passive galaxies found by a number of studies based on spectral indices (see [45] and references therein).