We present the results of a comprehensive re-analysis of the images of a vi
rtually complete sample of 28 powerful 3CR radio galaxies with redshifts 0.
6 < z < 1.8 from the Hubble Space Telescope (HST) archive. Using a two-dime
nsional modelling technique we have derived scalelengths and absolute magni
tudes for a total of 16 3CR galaxies with a median redshift of z=0.8. Our r
esults confirm the basic conclusions of Best, Longair & Rottgering in that
we also find z=1 3CR galaxies to be massive, well-evolved ellipticals, the
infrared emission of which is dominated by starlight. However, we in fact f
ind that the scalelength distribution of 3CR galaxies at z similar or equal
to 1 is completely indistinguishable from that derived for their low-redsh
ift counterparts from our own recently completed HST study of active galact
ic nuclei hosts at z similar or equal to 0.2. There is thus no evidence tha
t 3CR radio galaxies at z similar or equal to 1 are dynamically different f
rom 3CR galaxies at low redshift. Moreover, for a 10-object subsample we ha
ve determined the galaxy parameters with sufficient accuracy to demonstrate
, for the first time, that the z similar or equal to 1 3CR galaxies follow
a Kormendy relation that is indistinguishable from that displayed by low-re
dshift ellipticals if one allows for purely passive evolution. The implied
rather modest level of passive evolution since z similar or equal to 1 is c
onsistent with that predicted from spectrophotometric models provided one a
ssumes a high formation redshift (z greater than or equal to 4) within a lo
w-density universe. We conclude that there is no convincing evidence for si
gnificant dynamical evolution among 3CR galaxies in the redshift interval 0
< z < 1, and that simple passive evolution remains an acceptable interpret
ation of the K-z relation for powerful radio galaxies.