Children with autism spectrum disorders (ASD) adopt less efficient strategies than typically developing controls (TD) on verbal problem-solving tests such as the Twenty Questions Task. This study examined the hypotheses that this can be explained by differences in (i) planning processes or (ii) selective attention. Twenty-two children with ASD and 21 TD controls matched for age (Mage = 13:7) and cognitive ability (MFSIQ = 96.42) were tested on an adapted version of Twenty Questions and two planning tasks. ASD participants could recognize effective questions as well as TD participants on a forced-choice question discrimination task, but were observed to construct plans that were significantly less efficient. ASD performance was also specifically reduced when items could not be physically removed from the testing array, although this effect could be ameliorated by keeping a written record of participant questions during search. These findings indicate that ASD participants are sensitive to the within-task executive demands of Twenty Questions, but that their inefficiency in strategy relates to planning processes and question selection pretask. The implications for understanding ASD problem-solving skills and their impact on everyday functioning are discussed. Autism Res2011,4:401–411. © 2011 International Society for Autism Research, Wiley Periodicals, Inc.