<jats:title>Abstract</jats:title> <jats:p>In dynamic gravimetry, i.e. airborne and shipborne gravimetry, levelling methods are used to refine gravity disturbance results based on neighbouring trajectories. In the traditional crossover adjustment, line biases are estimated using gravity disturbance residuals at trajectory line crossings as input to a least-squares adjustment. In an alternative method, the results along the complete trajectory are used to estimate the gravity disturbance field in the survey area and line biases in a one-step least-squares adjustment applying spherical radial basis functions. This makes the bias estimation more robust since the observations are not restricted to a small number of residuals at crossings strongly affected by random errors. Adjustment becomes applicable to a wider range of campaigns including irregular trajectories without many crossings. Within the scope of this work, existing methods that estimate line biases are extended to bias estimation based on trajectory segments with inter-bias interpolation. The extended method can be particularly useful for irregular trajectories without a sufficient number of line crossings. The introduced levelling methods are evaluated at the example of three airborne campaigns: a fixed wing survey at Germany with a very dense grid, a fixed wing survey in East Antarctica with varying line separation, and a helicopter survey on Svalbard with highly irregular trajectories. It is shown that the levelling method based on spherical radial basis functions improves the precision in all evaluated campaigns, even when a traditional crossover levelling is not possible.</jats:p>