This study explored characteristics of children with autism with large intervention effects following a peer-mediated pragmatic language intervention, to devise algorithms for predicting children most likely to benefit. Children attended a 10-week intervention with a typically-developing peer. Data from a pilot study and RCT formed the dataset for this study. The POM-2 measured intervention outcomes. Children completed the EVT-2, TACL-4, and Social Emotional Evaluation at baseline, and parents completed the CCC-2 and CCBRS. High CCC-2 Use of Context and CCBRS Separation Anxiety scores and comparatively lower EVT-2, CCC-2 Nonverbal Communication and Cohesion scores predicted children with large intervention effects. Results can be used by clinicians to predict which children within their clinics might benefit most from participating in this intervention.