In several meta-analyses, self-estimates of abilities have been shown to correlate surprisingly low with individuals’ real (i.e., psychometrically assessed) abilities. We recently confirmed this in a study where we investigated the accuracy of self- and peer-estimates of six central abilities (verbal, numerical, spatial intelligence, interpersonal and intrapersonal competence, creative/divergent thinking). Here, we describe two studies: In study 1, we first investigated, to which extent self-estimates of adolescents’ central abilities can be predicted from three sources: relevant school grades, the pertinent psychometric ability itself, and personality (big five traits and narcissism). We found that self-estimates are a stronger reflection of the individuals’ personality than their abilities per se. Second, we wanted to assess to what degree (professional) interests, which might guide career decisions in adolescents/young adults, are predicted by self-estimated and psychometrically assessed abilities. We found that professional interests are mostly a function of self-estimates and not of ‘true’ abilities, a finding that we replicated in study 2 with young adults. Given the strong associations between self-estimates and personality and past findings showing that abilities are better predictors of professional success than personality traits are, this might be non-optimal.