We examined spatial variation in otolith geochemistry as a natural tag in juvenile Atlantic cod (Gadus morhua) to resolve geographic patterns during early life history. Individuals from 54 inshore sites spanned five embayments in eastern Newfoundland. Otolith composition differed at all spatial scales and related inversely to spatial scale. Classification analysis revealed increasing discrimination at coarser spatial scales: site (26%–58%), bay (49%), and coast (76%). Assignment success declined by ?10% per added site with increasing sampling sites per bay, demonstrating fine-scale (<100 km) variation. When we partitioned environmental variability from observed otolith chemistry using predictive models, assignment success improved by 56%, 14%, and 5% for site, bay, and coast, respectively. Our results demonstrate environmental influence on spatial structure of otolith chemistry and illustrate the importance of resolving baseline variability in otolith chemistry when conducting assignment tests. Collectively, our results describe the potential utility of juvenile otolith composition in evaluating contributions of subpopulations to the Northwest Atlantic cod stock and highlight important limitations imposed by environmental variation at scales less than 100 km.