As well as being considered a reliable measurement instrument of animal anxiety-like behavior, the elevated plus-maze (EPM) is also used as a post-hoc test to evaluate emotionality in genetically modified rodents. The present review considers factors which may further improve the validity (predictive/face/construct) of the EPM model: (1) the importance of measuring defensive patterns of response such as risk assessment in addition to traditional measures such as open arm time; (2) other methodological refinements such as min-by-min scoring and use of a test/retest protocol; and (3) the identification and control of major sources of variability in this test. To estimate whether current use of the EPM by researchers takes the above factor into account, a survey of the recent literature was conducted. Results showed that the majority of studies have not yet assimilated these important considerations into their use of the EPM. For example, although risk assessment measures may be more sensitive to anxiety modulating drugs than traditional measures, only a quarter of studies have adopted them. It is hoped that this review can provide insights into the optimal use of the EPM, a simple task that can be very complex in terms of behavioral analysis.