We present measurements of the normalised redshift-space three-point correlation function (Q_z) of galaxies from the Sloan Digital Sky Survey (SDSS) main galaxy sample. We have applied our "npt" algorithm to both a volume-limited (36738 galaxies) and magnitude-limited sample (134741 galaxies) of SDSS galaxies, and find consistent results between the two samples, thus confirming the weak luminosity dependence of Q_z recently seen by other authors. We compare our results to other Q_z measurements in the literature and find it to be consistent within the full jack-knife error estimates. However, we find these errors are significantly increased by the presence of the ``Sloan Great Wall'' (at z ~ 0.08) within these two SDSS datasets, which changes the 3-point correlation function (3PCF) by 70% on large scales (s>=10h^-1 Mpc). If we exclude this supercluster, our observed Q_z is in better agreement with that obtained from the 2dFGRS by other authors, thus demonstrating the sensitivity of these higher-order correlation functions to large-scale structures in the Universe. This analysis highlights that the SDSS datasets used here are not ``fair samples'' of the Universe for the estimation of higher-order clustering statistics and larger volumes are required. We study the shape-dependence of Q_z(s,q,theta) as one expects this measurement to depend on scale if the large scale structure in the Universe has grown via gravitational instability from Gaussian initial conditions. On small scales (s <= 6h^-1 Mpc), we see some evidence for shape-dependence in Q_z, but at present our measurements are consistent with a constant within the errors (Q_z ~ 0.75 +/- 0.05). On scales >10h^-1 Mpc, we see considerable shape-dependence in Q_z.