To evaluate the interrater and test-retest reliability, standard error of measurement (SEM), and the smallest detectable difference (SDD) of the Hand Assessment for Infants (HAI). HAI assessments of 55 infants (26 females, 29 males), 25 with clinical signs of unilateral cerebral palsy (CP) and 30 typically developing (mean [SD] age 6.8mo [2.4], range 3-11mo), were scored individually by three therapists. Three clinically experienced occupational therapists (OT 1-OT 3) with extensive experience in using the HAI, independently scored the video recorded HAI play sessions. Analysis of the combined group of infants and just the infants with clinical signs of unilateral CP (12 females, 13 males; mean age 7.6mo [2.1]) were conducted. Intraclass correlation coefficients (ICC, 2.1), Bland-Altman plots, SEM, and SDD were calculated. Interrater and test-retest reliability were excellent for the Both Hands Measure (BoHM) and the Each Hand Sum score (EaHS), with ICCs of 0.96 to 0.99. For individual items, the interrater and test-retest reliability was good to excellent (ICC 0.81-0.99). The SDD for the EaHS was 2 points, and for the BoHM the SDD it was 3 HAI units for infants with signs of unilateral CP. The HAI results showed good to excellent reliability. The SDDs were low, indicating that results beyond these levels exceed the measurement error and, thus, can be considered true changes. What this study adds The Hand Assessment for Infants (HAI) shows excellent reliability. A change of ≥3 HAI units is considered a true change. The HAI yields reliable measures for research and clinical practice.