Although directionality in interpreting has attracted considerable scholarly interest, to date, there are no observational studies providing empirical evidence on the directionality effect in the performance of professional interpreters who regularly work in both directions. Our aim is to partially fill this gap by exploring the relationship between disfluencies and directionality in simultaneous interpreting. To this end, we compare A-B and B/C-A interpretations from plenary debates of the European Parliament. Using English and Polish subcorpora extracted from the EP-Poland corpus, we performed quantitative analyses focusing on three types of disfluencies: anomalous pauses, hesitation markers, and false starts. Our initial assumption was that, in line with the prevalent belief, cognitive load is lower for into-A interpreting. Accordingly, we hypothesised that into-B interpretations should exhibit significantly more disfluencies. However, we have found no directionality effects related to disfluencies. Therefore, our results do not support the advantage of into-A interpreting. This finding raises some doubts about the legitimacy of favouring interpreting into the native language to the extent it is done now in many institutions.