The consequences of non-union of the scaphoid with or without deformity were evaluated before and after surgery with non-parametric tests, MANOVA and discriminant analysis, and Spearman correlation and contingency tables. 18 consecutive cases of scaphoid non-union, operated upon through a dorsal approach preserving the carpal ligaments since 1987, were used for measurements. Carpal height, radio-lunate and radio-scaphoid angles, lunate covering ratio, and apparent translation of the capitate and lunate were analyzed on standard views of the wrist before and after scaphoid repair. We found that the lunate covering ratio combined with the radio-lunate angle was the main indicator of deformity before the repair as suggested by the strong correlation between the radio-lunate angle and lunate covering ratio ( P= 0.0001). After repair, the only indicator of mal-union was the apparent translation of the lunate on frontal views (P= 0.0033) as suggested by the strong correlation between real-union and lunate translation ( P= 0.0001). Translation of both the lunate and capitate correlated well before and after repair. Radio-scaphoid angle and carpal height showed no significant changes and are not informative in the presence of subtle carpal alteration. Repair of the scaphoid produced an improvement in the radio-lunate angle in all cases, even in the presence of mal-union. However, full restoration was never achieved even in the absence of mal-union. More precise techniques of measurement might contribute to a better understanding of the carpal mechanics in relation to wefi-defined diagnoses such as scaphoid non-union.