We analyse parallel N-body simulations of three cold dark matter (CDM) univ
erses to study the abundance and clustering of galaxy clusters. The simulat
ion boxes are 500 h(-1) Mpc on a side and cover a volume comparable to that
of the forthcoming Sloan Digital Sky Survey. The use of a treecode algorit
hm and 47 million particles allows us at the same time to achieve high mass
and force resolution. We are thus able to make robust measurements of clus
ter properties with good number statistics up to a redshift larger than uni
ty. We extract haloes using two independent, public domain group finders de
signed to identify virialized objects - 'Friends-of-Friends' and 'HOP' - an
d find consistent results. The correlation function of clusters as a functi
on of mass in the simulations is in very good agreement with a simple analy
tic prescription based upon a Lagrangian biasing scheme developed by Mo & W
hite and the Press-Schechter (PS) formalism for the mass function. The corr
elation length of clusters as a function of their number density, the R-0-D
-c relation, is in good agreement with the APM Cluster Survey in our open C
DM model. The critical-density CDM model (SCDM) shows much smaller correlat
ion lengths than are observed. We also find that the correlation length doe
s not grow as rapidly with cluster separation in any of the simulations as
suggested by the analysis of very rich Abell clusters. Our SCDM simulation
shows a robust deviation in the shape and evolution of the mass function wh
en compared with that predicted by the PS formalism, Critical models with a
low as normalization or small shape parameter Gamma have an excess of mass
ive clusters compared with the PS prediction. When cluster-normalized, the
SCDM universe at z = 1 contains 10 times more clusters with temperatures gr
eater than 7 keV, compared with the PS prediction. The agreement between th
e analytic and N-body mass functions can be improved, for clusters hotter t
han 3 keV in the critical-density SCDM model, if the value of delta(c) (the
extrapolated linear theory threshold for collapse) is revised to be delta(
c)(z)= 1,686[(0.7/sigma(8))(1 + z)](-0.125) (sigma(8) is the rms density fl
uctuation in spheres of radius 8 h(-1) Mpc). Our best estimate for the ampl
itude of fluctuations inferred from the local cluster abundance for the SCD
M model is sigma(8) = 0.5 +/- 0.04. However, the discrepancy between the te
mperature function predicted in a critical-density universe and that observ
ed at z = 0.33 (Henry et al.) is reduced by a modest amount using the modif
ied PS scheme. The discrepancy is still large enough to rule out Omega(0) =
1, unless there are significant differences in the relation between mass a
nd temperature for clusters at high and low redshift.