This study explored the reliability and discriminant validity of a new narrative measure, the Teddy Bear's Picnic (Mueller, 1996), in a small non-clinical community sample of 5 year olds (n=35). The measure is designed for use with young children, and reveals internally represented psychosocial themes through the use of a story telling technique. The Teddy Bear's Picnic measure consists of nine incomplete story stems which are administered to children individually, using a range of age appropriate toys and props. Each story is based on an imaginative scenario involving a family of toy bears. Children are presented with the stories in a specific order, each of which poses an unresolved hypothetical conflict; following which they are asked to complete the unfinished story in response to the prompt "what happens next? " Concurrent parent and teacher measures were gathered as part of a larger ongoing study in the same community. Parents and teachers were asked to complete a brief 25 item rating scale, the Strengths and Difficulties Questionnaire (SDQ: Goodman, 1997), and one categorical question asking them to rate level of concern about the child's behaviour (based on the work of Stallard, 1995). The study aimed to examine the associations between coded thematic material generated from the Teddy Bear's Picnic, and parent and teacher rated behaviour problems and global concern about behaviour. Associations were all in the expected direction, but not statistically significant. Inter-rater reliability was acceptable for total positive theme scores (r=. 91, p<. 001), and approached acceptability for negative total theme scores(r=. 58, p<. 05). Problems were particularly apparent in the use of TBP composite total scores. Internal consistency of total theme scores and factor based subscale scores (derived from Mueller, 1996), revealed low alpha coefficients (alpha = . 42 -. 59). Discriminant function analysis demonstrated that in this study, the TBP was unable to correctly classify children into parent and teacher rated groupings, based on both measures. Further studies would benefit from larger samples than were available to Mueller (1996) and the present study. However, the measure appears promising as a method for eliciting psychologically relevant themes from young children, but findings suggest that further work is required in developing it's psychometric properties.