The role of computer training has long been critical in organizations as reliance on technology for strategic advantage increases in importance. How to most effectively conduct such training has clear implications for organizations. This study examines one area of training which is not well understood: the role that competence level plays in the self-efficacy-competence relationship (if indeed it plays a role at all) during skill acquisition. Two opposing conceptual positions are presented from the literature, one that suggests the relationship between self-efficacy and competence will be stronger early in the skill acquisition process (when competence is minimal), the other suggesting the strength of the relationship will be stronger at mastery. Using a sample of over 600 and structural equation modeling, the relationship between self-efficacy and competence for six different computing application domains is tested by dividing respondents in each domain in half, according to competence level. Results empirically demonstrate that level of competence makes a significant difference in the domains, that those higher in ability typically have a stronger relationship with self-efficacy. Results also show that the relationship is weaker for those new to the application and those who have mastered the application. The important implications for training are discussed.