Aim: The cognitive-linguistic abilities of 2 subgroups of children with speech impairment were compared to better understand underlying deficits that might influence effective intervention. Methods: Two groups of 23 children, aged 3;3 to 5;6, performed executive function tasks assessing cognitive flexibility and nonverbal rule abstraction. Following the system of differential diagnosis of speech disorders first described by Dodd, Leahy, and Hambly (1989), one group was identified as having delayed speech development, as their non-age-appropriate speech error patterns were typical of younger children. The other group was diagnosed as disordered because children consistently used at least one speech error pattern atypical of any age group in an assessments' normative sample (Dodd, Zhu, Crosbie, Holm, & Ozanne, 2002). Results and Conclusions: The disordered group performed less well than the delayed group: They had poorer cognitive flexibility and difficulty abstracting nonlinguistic rules. They made more consonant errors and different types of errors. The 2 groups did not differ on measures of language, vowel accuracy, or consistency of multiple productions of the same words. The findings suggest that different interventions, reflecting knowledge of underlying deficits, might benefit specific subgroups of children with speech impairment.