Purpose: The aim was to evaluate the effectiveness of computer-assisted input-based intervention for children with speech sound disorders (SSD). Method: The Sound Start Study was a cluster-randomized controlled trial. Seventy-nine early childhood centers were invited to participate, 45 were recruited, and 1,205 parents and educators of 4- and 5-year-old children returned questionnaires. Children whose parents and educators had concerns about speech were assessed (n = 275); 132 children who were identified with phonological pattern-based errors underwent additional assessment. Children with SSD and no difficulties with receptive language or hearing, typical nonverbal intelligence, and English as their primary language were eligible; 123 were randomized into two groups (intervention n = 65; control n = 58), and 3 withdrew. The intervention group involved Phoneme Factory Sound Sorter software (Wren & Roulstone, 2013) administered by educators over 9 weeks; the control group involved typical classroom practices. Participants were reassessed twice by a speech-language pathologist who was unaware of the initial assessment and intervention conditions. Results: For the primary outcome variable (percentage of consonants correct), the significant mean change from pre- to postintervention for the intervention group (mean change = +6.15, p < .001) was comparable in magnitude to the significant change for the control group (mean change = +5.43, p < 0.001) with a small between-groups effect size for change (Cohen's d = 0.08). Similar results occurred for measures of emergent literacy, phonological processing, participation, and well-being. Conclusion: Computer-assisted input-based intervention administered by educators did not result in greater improvement than typical classroom practices.