ABSTRACT Close pairs of galaxies have been broadly studied in the literature as a way to understand galaxy interactions and mergers. In observations, they are usually defined by setting a maximum separation in the sky and in velocity along the line of sight, and finding galaxies within these ranges. However, when observing the sky, projection effects can affect the results, by creating spurious pairs that are not close in physical distance. In this work, we mimic these observational techniques to find pairs in the three hundred simulations of clusters of galaxies. The galaxies’ 3D coordinates are projected into 2D, with Hubble flow included for their line-of-sight velocities. The pairs found are classified into ‘good’ or ‘bad’ depending on whether their 3D separations are within the 2D spatial limit or not. We find that the fraction of good pairs can be between 30 and 60 per cent depending on the thresholds used in observations. Studying the ratios of observable properties between the pair member galaxies, we find that the likelihood of a pair being ‘good’ can be increased by around 40, 20, and 30 per cent if the given pair has, respectively, a mass ratio below 0.2, metallicity ratio above 0.8, or colour ratio below 0.8. Moreover, shape and stellar-to-halo mass ratios, respectively, below 0.4 and 0.2 can increase the likelihood by 50 to 100 per cent. These results suggest that these properties can be used to increase the chance of finding good pairs in observations of galaxy clusters and their environment.