In a previous study a group of children with cerebral palsy (CP) were found to have considerable difficulties with narratives, performing several standard deviations below the criteria for the Information score of the Bus Story Test (BST). To examine in depth the performance of children with CP and a control group with typically developing (TD) children on a narrative task, in order to search for possible underlying causes to the problems in the CP group. The results of the BST for 10 children with CP, mean age 7;11 years, were investigated. The analysis of the BST was supplemented with the use of the Narrative Assessment Profile (NAP) and quantitative analyses of number of words, mazes, propositions, types of conjunctions and story elements. A significant relationship between the explicitness dimension on the Narrative Assessment Profile and the BST Information score in the CP group suggested that the problems could be derived to a limited use of cohesion and a scarcity of essential information. Compared to the CP group, the TD group used significantly more causal conjunctions. The results indicate a general problem with cohesion at the textual level in the CP group. A further finding was the occurrence of a positive correlation between the use of mazes and the BST Information score in the CP group. These results have implications for the design of a more specific intervention for children, where the NAP was found to be a valuable tool in combination with the BST or other assessment materials. Further, it is shown that mazes, mostly regarded as a behaviour that not enhances speech production, for some children can be used as a means to find necessary words and pieces of information. (Contains 3 tables and 1 figure.)