A computerised teaching tool known as the T-IRAP has been previously shown to be an effective tool for teaching relational responding skills to children with autism, using a multiple baseline design to demonstrate greater speed and accuracy with the T-IRAP compared to Table-top procedures (Kilroe, Murphy, & Barnes-Holmes, 2011).The current study extends this by using an alternating treatments design which is not subject to problematic sequence effects that may result with multiple baseline design. Cassidy, Roche, & Hayes, (2011) found that a teaching complex forms of relational responding resulted in significant gains in cognitive ability as measured by IQ tests. The current study builds upon this by examining the IQ and verbal abilities of children with autism at pre and post test to determine if relational training will result in similar increases in IQ. Pre and post intervention measures of IQ and verbal ability were taken to assess if relational training generates any gains on these measures. An alternating treatments design compared a Table Top (TT) procedure with a computerised procedure (T-IRAP) to teach the same sets of relational stimuli. Study 1 successfully taught five children flexible relational responding with simple non-arbitrary stimuli. Study 2 successfully taught flexible relational responding with arbitrary stimuli to one participant. Finally, Study 3 taught complex relational flexibility with double contingency reversals to five participants with diagnosed autism. All participants demonstrated some degree of relational flexibility, with a range of competencies observed. T-IRAP was found to be more efficient and effective in terms of speed and accuracy as compared to TT teaching across the research program as a whole. No statistically significant differences were found in ability scores from pre to post-test assessment, some individual gains were observed however.