Objective: To elucidate whether motor deficit severity mediates effects of intellectual function in predicting motor speech functions in children with cerebral palsy (CP).Method: Twenty-five children with spastic CP aged 5-8 were classified into two groups, levels I-II (n=13) and levels III-IV (n=12) based on Gross Motor Function Classification System (GMFCS) levels. Predictor was the intelligence quotient (IQ). Speech outcome measures were Percentage of Consonants Correct (PCC) and Verbal Motor Production Assessment for Children (VMPAC). Pearson correlation analysis was used to investigate relationships between IQ and speech outcomes. A mediation analysis (a three-variable path model) was used to explain how GMFCS levels (the mediator) affect IQ to predict speech outcome.Results: Children with GMFCS levels I-II had greater scores in all IQ and all VMPAC sub scores, and PCC than those with GMFCS levels III-IV (p<0.05). Pearson correlation analysis showed that full IQ (FIQ) was correlated with all VMPAC scores (r=0.42-0.62, p<0.05), but not with PCC scores. The path model assumes path a from the predictor variable (FIQ) to the mediator (GMFCS levels), path b from the impact of the mediator (GMFCS levels) on the outcome variable (VMPAC), and two causal paths to the outcome variable (paths c and c'). Mediation analysis revealed that GMFCS level completely mediated the effect of FIQ when predicting global motor control (a=-0.027 ± 0.01, p=0.01; b=-17.910 ± 3.620, p<0.001; c=0.537 ± 0.235, p=0.032; c'=0.058 ± 0.192, p=0.766), focal motor control (a=-0.027 ± 0.01, p=0.01; b=-9.287 ± 2.441, p=0.001; c=0.315 ± 0.140, p=0.035; c'=0.067 ± 0.129, p=0.611), and average VMPAC (a=-0.027 ± 0.01, p=0.01; b=-10.976 ± 2.521, p<0.001; c=0.455 ± 0.154, p=0.007; c'=0.161 ± 0.133, p=0.241). GMFCS level partially mediated the effect of FIQ when predicting the VMPAC sequence (a=-0.027 ± 0.01, p=0.01; b=-5.732 ± 2.767, p=0.050; c=0.512 ± 0.135, p=0.001; c'=0.359 ± 0.146, p=0.023). Conclusion:Motor deficit severity mediates effect of intellectual functions on speech motor control. The findings allow clinicians to identify early motor speech problems in children with CP using the GMFCS level alone.