Abstract Aim To expand the scope of the Hand Assessment for Infants (HAI) by validating its internal scale structure and reliability in infants with a corrected age of 3 to 12 months who are at risk of bilateral or unilateral cerebral palsy (CP). Method This test–validation study included 274 HAI assessments collected from 221 infants (mean age = 7 months, SD = 2.4 months, 132 males) at risk of unilateral ( n = 123) or bilateral ( n = 98) CP. Each assessment was scored on 17 HAI items (12 unimanual and five bimanual) and analysed using the Rasch measurement model. Results Strong internal scale validity and person and item reliability were found for the HAI items when analysed separately for infants at risk of unilateral and bilateral CP. Logit measures for both scales were linked to a common reference frame, allowing comparable overall bimanual performance measures (HAI unit) while maintaining distinct item difficulty hierarchies. Interpretation The HAI provides valid and reliable measures of hand use in infants at risk of bilateral and unilateral CP regardless of the severity of manual impairments. By linking the scales, the HAI unit provides a comparable measure of bimanual performance across CP subtypes. Thus, the subtype of CP does not need to be determined before data collection.