Safety interventions must be approached in a holistic manner, by taking cognisance of the organisational system in which accidents present themselves. Thus, a need to understand the individual in context of the organisational system that he/she functions in exists. This study focuses on the relationship between attitudes, beliefs, intention, perception, control and the propensity of employees to engage in unsafe behaviour, with a view towards improving safety statistics. In this regard the study investigated the construct validity of a Safety Survey instrument, based on the responses (n=450) of employees in the mining industry. An Exploratory Factor Analysis showed that the Safety Survey instrument differentiated between two to three underlying factors. The result of the Confirmatory Factor Analysis tested two models by demonstrating the minimum requirements of the goodness of fit indexes. The construct validity of the Safety Survey instrument could be established for two nested models that identified an internal and external locus of control factor. The statistical evidence indicated an acceptable model fit. The statistical evidence validated the construct validity of the measurement model. Copyright 2008, University of Pretoria. All rights reserved. The copyright in this work vests in the University of Pretoria. No part of this work may be reproduced or transmitted in any form or by any means, without the prior written permission of the University of Pretoria. Please cite as follows: Windbacher, M 2008, Determining the construct validity of the safety survey instrument, MCom dissertation, University of Pretoria, Pretoria, viewed yymmdd < http://upetd.up.ac.za/thesis/available/etd-02202012-123434 / > C12/4/66/gm / Dissertation (MCom)--University of Pretoria, 2008. / Human Resource Management / unrestricted