Introduction: The objective of the current study was to design and pilot-test a nonverbal intelligence test for children and adolescents with visual impairment (i.e., blindness and low vision) aged 10 to 16 years. Method: A three-dimensional haptic matrix test of nonverbal cognitive abilities (3-DHMT-CA), initially developed for adults with visual impairment, was adapted and administered to 25 juveniles with severe low vision or blindness and 25 sighted peers. The test consisted of 22 tactual matrices with a missing element, and juveniles were asked to understand the logical scheme underlying each matrix and to complete it. Convergent validity was assessed in the sighted group with the Standard Progressive Matrices. Divergent validity was measured with the Verbal Comprehension Index (VCI) and the Working Memory Index (WMI) of the Italian version of the Wechsler Intelligence Scale for Children–Fourth Edition. Between-group differences were analyzed with Student’s t-test and ANOVA. Results: Results show a high correlation with the convergent measure (with the Standard Progressive Matrices), and the moderate correlations with divergent measures (VCI and WMI). Between-group differences showed significantly higher scores for sighted peers than juveniles with visual impairment at 3-DHMT-CA, and non-significant differences at VCI and WMI. Discussion: This haptic nonverbal cognitive test showed good psychometric properties, but it needs to be validated in a larger sample with a qualitative analysis of solution strategies because juveniles with visual impairment seem to use different verbal and working memory strategies. Implications for Practitioners: The 3-DHMT-CA and the haptic two-hands modality seem to be suitable to test nonverbal cognitive abilities in juveniles with visual impairment. Practitioners should focus on empowering verbal and working memory abilities, since they may affect nonverbal cognitive abilities as well.