The Knowledge Capital Model (KC-model), described in Markusen (2002), encompasses both market size (horizontal) as well as factor endowment (vertical) explanations to why multinational production occurs. Although the KC-model seems intuitively appealing, the empirical support has, so far, been weak and even confused. In this study, we find strong, robust and consistent support for the KC-model. In contrast to previous studies, our skill measures follow directly from the model. We also use an enlarged dataset, where the date coverage is significantly improved. Our results also give estimated surfaces remarkably similar to theoretical simulations of the KC-model. In addition, the results give important insights into why previous studies have yielded such diverse results.