Structured 'communicative temptation' procedures administered by a clinician were compared to unstructured parent-child interactions in sampling intentional communicative acts in 11 subjects with intellectual disability. The subjects were assessed twice over a 5-7 month period. The results indicated that the structured condition was more effective in sampling Requests and Comments, with more Requests than Comments produced. The unstructured condition was more effective in sampling responses, but only during the second assessment. Although there was no difference in the total number of intentional communicative acts produced across assessment times, there was an increase in the use of linguistic forms during the second assessment: that is, the subjects used more speech and signs and fewer gestures and general vocalisations during the second assessment than the first. The findings suggest the usefulness of combining structured and unstructured conditions in providing information on the variety of children's communicative acts and the linguistic level at which these are expressed.