A new technique removes the leveling errors in airborne geophysical data based on an assumption that the data are continuous (i.e., not renulled or recalibrated) from line to line. A single flight line is selected as a reference to level and tie all survey lines to this continuously varying datum. The leveling errors are determined in a least-squares sense from the reference line and the adjacent line to be leveled. The technique markedly improves the quality of the unleveled, raw data and works best if the reference line overlaps most of the line being leveled. This method cannot distinguish a linear geologic feature in the direction of flight lines from a leveling error. The technique initially was developed for airborne electromagnetic (EM) data in cases where tie-line leveling works poorly. However, it may also be used for airborne-magnetic and other data, in which case tie-lines used in conventional leveling techniques are not needed.