This study investigates the effects of teacher sex and pupil sex in the teachers' perception of the seriousness of three extracted patterns of pupils' undesirable behaviours. In contraposition to Langfeldt's (1992) conclusion that gender differences are no longer observable when investigating complex patterns of behaviour as opposed to single behavioural items the present writers argue that where gender is a true effect this will also be evident at a factorial level of abstraction. A sample of 844 primary school teachers in Malta rated the seriousness of 16 selected undesirable behaviours when these are manifest in boy pupils and in girl pupils. A principal components analysis revealed three patterns of perceived problem behaviours labelled 'withdrawal behaviour', 'dissocial/overtly challenging behaviour' and 'disruptive behaviour'. Gender differences were found in each of the three factors: the main effect pupil sex was significant on 'withdrawal behaviour' and 'disruptive behaviour', and teacher sex on 'dissocial/overtly challenging behaviour'. Results also indicated a significant interaction effect involving teacher sex and pupil sex on the factor 'withdrawal behaviour'.