This paper decomposes the variance in EU food industry return‐on‐assets into year, country, industry and firm effects using a hierarchical linear model (HLM). The HLM approach accounts for some of the methodological drawbacks of conventional approaches of variance decomposition such as anova and components of variance and additionally allows the estimation of the impact of covariates within each effect level. The results for selected EU countries show that firm effects are far more important than industry structure in determining food industry profitability. In particular, firm size and industry concentration are drivers of profitability while firm risk and age as well as industry growth have a negative influence.