Heterogeneity effects are compared in a power reactor subassembly of the Superphénix type and in the lattices of the critical experiments performed in the Masurca critical facility. Both the fuel pin heterogeneity and the structure tube heterogeneity are evaluated with a two-step method based on the subgroup technique for self-shielding effect evaluation and on the Benoist method for streaming effect evaluation (the DHARMA method). Besides validation with reference calculations for simple geometries, experimental evidence confirms the good performance of the method proposed. It is found that the components of the heterogeneity effects on keff play a different role in power reactors and critical experiments, and this fact has a significant impact in bias factor assessment or cross-section adjustments based on integral experiments.