Many leadership development programs assume that m anagers utilize feedback from subordinates to prepare a personal improvement agenda. Encouraging the use of such feedback assumes that (a) raters can clearly distinguish between different dimensions of leadership behavior, (b) that the data reflect managers’ personal leadership style or predispositions rather than other, external influences, and (c) that the rated behaviors are valued by the organization. This study examines the validity of these assumptions in a sample of engineering managers undergoing leadership training. Results indicate that although subordinates did make distinctions between different leadership dimensions, organizational roles of the leaders were significantly related to such ratings, and organizational rewards were significantly but weakly related to only one of five leadership dimensions being assessed. Results are discussed in terms of their implications for leaders and organizations trying to make sense of subordinate data in leadership development contexts.