Geometric morphometrics is a relatively new tool to fisheries research showing promise as a means of enabling researchers to cheaply and quickly categorise fish to individual stocks based on variations in otolith form, most commonly size and shape. In this study we introduce the method of elliptical Fourier analysis using two widely separated populations of striped trumpeter (Latris lineata) as a case study and compare the interpretation of results based on both unconstrained and constrained ordination techniques. There were no significant differences in otolith morphometrics between sex or age classes within each region. All form descriptors were standardised for fish length, thereby minimising confounding effects on any potential inter-regional otolith form differences. Non-metric multidimensional scaling was not sufficient to elucidate differences in otolith form between populations. However, using constrained canonical analysis of principal coordinates and canonical discriminant analysis, regional differences became evident with allocation success of 75 and 87%, respectively. Based on this study differences in otolith form reflect that the two tested striped trumpeter populations have reasonable phenotypic anonymity. This study further supports the usefulness of shape analysis and constrained non-parametric statistical tests as tools for stock discrimination and introduces elliptical Fourier analysis to the study of otolith morphometrics.