Abstract The effectiveness of discrimination of phenotypic stocks was compared between otolith and scale shapes for Mugil curema specimens collected at five different locations in the Gulf of Mexico and two locations along the Pacific coast during two consecutive years. Geometric morphometric methods were used to determine the discrimination among locations using seven and 22 landmarks for scales and otoliths, respectively. The cross-validated discriminant analysis by location correctly classified 43.2 and 40.2% based on shape variables (Principal Components scores) for otoliths for all locations jointly, while for scales the classification percentages were 48.7 and 47.4% for the first and second years, respectively. Classification results improved when the discrimination analyses were carried out for pairs of locations, with 51.4 to 82.6% for otoliths and 72.7 to 97.1% for scales. The analysis was run for two consecutive years and the results for both years were best for the scales. Thus, fish scale shape offers a straightforward, non-destructive, accessible, quick and inexpensive method to trace fish phenotypic stocks.