Background: Predictive validity studies for selection criteria into graduate entry courses in Australia have been inconsistent in their outcomes. One of the reasons for this inconsistency may have been failure to have adequately considered background disciplines of the graduates as well as other potential confounding socio-demographic variables that may influence academic performance. Methods. Graduate entrants into the MBBS at The University of Western Australia between 2005 and 2012 were studied (N = 421). They undertook a 6-month bridging course, before joining the undergraduate-entry students for Years 3 through 6 of the medical course. Students were selected using their undergraduate Grade Point Average (GPA), Graduate Australian Medical School Admissions Test scores (GAMSAT) and a score from a standardised interview. Students could apply from any background discipline and could also be selected through an alternative rural entry pathway again utilising these 3 entry scores. Entry scores, together with age, gender, discipline background, rural entry status and a socioeconomic indicator were entered into linear regression models to determine the relative influence of each predictor on subsequent academic performance in the course. Results: Background discipline, age, gender and selection through the rural pathway were variously related to each of the 3 entry criteria. Their subsequent inclusion in linear regression models identified GPA at entry, being from a health/allied health background and total GAMSAT score as consistent independent predictors of stronger academic performance as measured by the weighted average mark for the core units completed throughout the course. The Interview score only weakly predicted performance later in the course and mainly in clinically-based units. The association of total GAMSAT score with academic performance was predominantly dictated by the score in GAMSAT Section 3 (Reasoning in the biological and physical sciences) with Section 1 (Reasoning in the humanities and social sciences) and Section 2 (Written communication) also contributing either later or early in the course respectively. Being from a more disadvantaged socioeconomic background predicted weaker academic performance early in the course. Being an older student at entry or from a humanities background also predicted weaker academic performance. Conclusions: This study confirms that both GPA at entry and the GAMSAT score together predict outcomes not only in the early stages of a graduate-entry medical programme but throughout the course. It also indicates that a comprehensive evaluation of the predictive validity of GAMSAT scores, interview scores and undergraduate academic performance as valid selection processes for graduate entry into medical school needs to simultaneously consider the potential confounding influence of graduate discipline background and other socio-demographic factors on both the initial selection parameters themselves as well as subsequent academic performance. © 2014 Puddey and Mercer; licensee BioMed Central Ltd.