Summary: The paper derives analytical bias expressions for the least squares bearings-only target localization algorithms. A detailed analysis of the interplay between the target localization geometry and the estimation bias is provided for the cases of elliptical and linear observer trajectories. It is observed that the estimation bias is proportional to the distance between the target and the centre of mass of the observer positions. The effect of the localization geometry parameters on the estimation bias is illustrated analytically and by way of numerical examples. The theoretical findings of the paper are backed up with simulation studies.