In this paper, we empirically re-assess the question which theoretical models and motives are most suitable to explain global patterns of foreign direct investment (FDI). Compared to previous studies, we use bilateral FDI positions with a much more comprehensive coverage of emerging and developing economies, the IMF’s CDIS. We apply cross validation to assess the performance of the gravity model and the knowledge capital (KK) model and add cultural, institutional, and financial factors, as suggested by theories on FDI determinants. We find the gravity model to achieve the best theory-consistent out-of-sample prediction, particularly when parameter heterogeneity of South and North FDI is allowed for. Controlling for surrounding market potential is important to recover the horizontal effect of the gravity model. Including institutional, cultural, or financial factors does not improve the model performance distinctly although results for those variables are mostly in line with theory.