Introduction: Before 2015 the hand classification in wheelchair rugby consisted of non-sport specific tests. The hand classification was not in accordance with the classification code introduced by the International Paralympic Committee in 2003. In 2015, the newly revised wheelchair rugby classification manual was released, containing the revised wheelchair rugby hand classification. Hand tests that were not functional sport-specific tests were removed from the bench test in wheelchair rugby classification. Lumbrical, interossei and thumb opposition manual muscle testing were added to the bench test in wheelchair rugby classification. On both national and international levels of classification, classifiers verbalised their uncertainty to their fellow panel members regarding their hand placement on the athlete’s hand and interpretation of the manual muscle testing of the hand that was observed and tested. This justified reliability testing of the new hand classification. Aim: The aim of this study was to determine the intra and inter-rater reliability of the manual muscle testing in the new hand classification of wheelchair rugby. Study design: This study followed a quantitative non-experimental, cross-sectional design. Method: The raters who took part in the study were active international wheelchair rugby classifiers from all over the world. The raters received an electronic questionnaire consisting of biographic information and three videos repeated two times. Each video showed an athlete’s hand being classified by a classifier. The raters had to give a manual muscle test grade for each subject (muscle) tested in each video by using tick boxes. The manual muscle test grades that could be given were: 0-1, 2, 3 and 4-5. The first three raters in each international wheelchair rugby classification level who completed the questionnaire were used for the data analysis. Data Analysis: The statistician used the two way model for the ICC in which each subject was rated by the same raters to determine the absolute agreement for each objective. The Medcalc program was used. To indicate the strength of agreement the ranges provided by Landis and Koch (1977) were used: 0.0 – 0.2 slight, 0.21 – 0.4 fair, 0.41 – 0.6 moderate, 0.61 – 0.8 substantial and 0.81 – 1.00 almost perfect. Conclusion: Raters one, two, five, seven, eight and nine’s intraclass correlation coefficient values fell between 0.81-1.00 which is descriptive of almost perfect levels of intra-rater reliability. Raters three, four and six’s intraclass correlation coefficient values fell between 0.61-0.80 which is descriptive of substantial levels of intra-rater reliability. However, none of the raters scored 100% when accuracy was determined. All three levels had intraclass correlation coefficient values which is descriptive of almost perfect levels of intra-rater reliability within each level. Level 2, 3 and 4 classifiers had intraclass correlation coefficient values between 0.81- 1.00 which is descriptive of almost perfect levels of inter-rater reliability when the manual muscle testing grades for the first and repeated videos were compared. Across all nine raters there was a high intraclass correlation coefficient value which was descriptive of almost perfect inter-rater reliability. The accuracy in each level and across all nine raters was low. Finger extensors, thumb abductor and thumb flexor showed intraclass correlation coefficient values between 0.41-0.6 which is descriptive of moderate levels of intrarater reliability. The only subjects (muscles) that were graded accurately when compared to a memorandum were subjects with a manual muscle test grade 0-1 and 4-5. Most of the accurate manual muscle test grades were for athlete two in the video footage. Athlete two was classified as having a 2.0 hand. / Dissertation (MOccThe)--University of Pretoria, 2017. / Occupational Therapy / MOccThe / Unrestricted