ABSTRACT We explore the clustering of galaxy groups in the Galaxy and Mass Assembly (GAMA) survey to investigate the dependence of group bias and profile on separation scale and group mass. Due to the inherent uncertainty in estimating the group selection function, and hence the group autocorrelation function, we instead measure the projected galaxy–group cross-correlation function. We find that the group profile has a strong dependence on scale and group mass on scales $r_\bot \lesssim 1 h^{-1}$. We also find evidence that the most massive groups live in extended, overdense, structures. In the first application of marked clustering statistics to groups, we find that group-mass marked clustering peaks on scales comparable to the typical group radius of r⊥ ≈ 0.5 h−1. While massive galaxies are associated with massive groups, the marked statistics show no indication of galaxy mass segregation within groups. We show similar results from the IllustrisTNG simulations and the L-Galaxies model, although L-Galaxies shows an enhanced bias and galaxy mass dependence on small scales.