Particularly important is Schmidt & Hunter’s 1998 review paper, The validity and utility of selection methods in personnel psychology: Practical and theoretical implications of 85 years of research findings. A useful summary is here. In summary, they find GMA (General Mental Ability) to be the best predictor of job performance:
“GMA can be considered the primary personnel measure for hiring decisions and one can consider the remaining personnel measures as supplements to GMA measures”
They also find that supplementing GMA tests with structured interviews provides best combination of predictors.
Note that the useful process is that of structured interviews. The usual practice of unstructured interviews has a relatively poor predictive validity. It gets worse when you consider that its been consistently found that people make up their minds about their interlocutors in the first few seconds of interaction (Ambady & Rosenthal, 1993) , and that non verbal behaviour is a crucial determinant of job interview results (Levine and Feldman, 2002).