Purpose: The purpose of this paper was to assess two different leader development interventions by comparing their effects on leadership behaviour and evaluating their combined impact after two years, from the viewpoints of both the participating managers and external raters. Design/methodology/approach: The study was a longitudinal randomised controlled trial with a cross-over design. Health care managers (n = 177) were first randomised to either of two 10-month interventions and a year later were switched to the other intervention. Leadership behaviour was rated at pre-test and 12 and 24 months by participating managers and their superiors, colleagues and subordinates using a 360-degree instrument. Analysis of variance and multilevel regression analysis was performed. Findings: No difference in effect on leadership behaviour was found between the two interventions. The evaluation of the combined effect of the interventions on leadership behaviour showed inconsistent (i.e. both increased and decreased) ratings by the various rater sources. Practical implications: This study provides some evidence that participation in leadership development programmes can improve managers’ leadership behaviours, but the results also highlight the interpretive challenges connected with using a 360-degree instrument to evaluate such development. Originality/value: The longitudinal randomised controlled design and the large sample comprising both managers and external raters make this study unusually rigorous in the field of leadership development evaluations.