As a form of aptitude testing for interpreting, recall across languages has long been considered as a valid means of assessing candidates' linguistic, cognitive and communicative abilities. However, most of the existing research on this topic is only derived from the consensus among or the professional experience of interpreting researchers, practitioners and trainers. There is still a paucity of grounded research on the validity of the recall tests. Empirical research on this topic remains scant. To fill this gap, the present study adopts a data-driven approach to exploring the predictive validity of the recall tests as a whole as well as to what extent each of the four assessment parameters (fidelity, language, coherence and delivery) can predict the candidates' end-of-year interpreting performance in the Chinese context. Data analysis suggests that candidates' performance on the recall tests could predict their overall interpreting performance in both language directions, with the criterion of language as a more powerful predictor ([Beta][subscript 1] = 0.45) than the criterion of fidelity ([Beta][subscript 2] = 0.30) in the C-E recall and fidelity as a more powerful predictor ([Beta][subscript 1 = 0.44) than language ([Beta][subscript 2] = 0.31) in the E-C recall. Implications of the findings of this study were also discussed.