In this talk I will discuss issues relevant to the use of galaxy cluster abundances to constrain the properties of dark energy. In particular I will focus on the use of self-calibration to jointly constrain cosmological and cluster model parameters using the large cluster samples provided by SZ surveys. These surveys should have sufficient statistics and sensitivity to dark energy to remain competitive with other dark energy probes, provided the connection between cluster observables and mass is describable in a small number of extra nuisance parameters. I will discuss the distribution of cluster SZ observables extracted from a large sample of simulated clusters and the link between halo assembly history and scatter in the mass-observable relations.