Objective: The aims were to (i) provide a German translation of the Melbourne Assessment 2 (MA2), a quantitative test to measure unilateral upper limb function in children with neurological disabilities and (ii) to evaluate its reliability and aspects of clinical utility. Methods: After its translation into German and approval of the back translation by the original authors, the MA2 was performed and videotaped twice with 30 children with neuromotor disorders. For each participant, two raters scored the video of the first test for inter-rater reliability. To determine test–retest reliability, one rater additionally scored the video of the second test while the other rater repeated the scoring of the first video to evaluate intra-rater reliability. Time needed for rater training, test administration, and scoring was recorded. Results: The four subscale scores showed excellent intra-, inter-rater, and test–retest reliability with intraclass correlation coefficients of 0.90–1.00 (95%-confidence intervals 0.78–1.00). Score items revealed substantial to almost perfect intra-rater reliability (weighted kappa kw¼0.66–1.00) for the more affected side. Score item inter-rater and test–retest reliability of the same extremity were, with one exception, moderate to almost perfect (kw¼0.42–0.97; kw¼0.40–0.89). Furthermore, the MA2 was feasible and acceptable for patients and clinicians. Conclusions: The MA2 showed excellent subscale and moderate to almost perfect score item reliability.