The primary goal of the current study was to examine the ability of children with autism (ASD) and children with learning disabilities (LD) to improve their metaphoric competence by an intervention program using "thinking maps". Twenty ASD children, 20 LD, and 20 typically developed (TD) children were tested on metaphors and idioms comprehension tests, homophone meaning generation test, and fluency tests. Both ASD and LD groups performed poorly compared with TD on all tests, with the LD group outperformed the ASD group in the executive function tests. The results indicate that the LD group was able to use the "thinking maps" to understand metaphors that were encountered for the first time more efficiently than the ASD group. Furthermore, in the autistic group the homophone meaning generation test, associated with mental flexibility mechanism, correlated with novel metaphors understanding, which do not rely on prior knowledge. In the learning disabilities group, conventional metaphors understanding correlated with the homophone meaning generation test. (Contains 3 tables and 2 figures.)