A reliable estimate of the anoxic regions in soil is essential in predicting the activity of microbial processes such as denitrification. In this study, two models were tested that have been previously proposed to estimate anaerobiosis within saturated aggregates. The first model assumes O-2 consumption is constant as long as O-2 is present; in the second model, O-2 consumption depends on O-2 concentration. The test was based on the overall respiration of natural aggregates of different sizes placed at different O-2 concentrations. Some physical, chemical, and biological properties of the aggregates were also analyzed to verify the basic assumptions of the models. This analysis included direct measurements of O-2 concentration within aggregates. The second model better described aggregate respiration, but some discrepancies were found between data and theory. Oxygen profiles indicate a steep fall of O-2 near the aggregate surface. This suggests that the potential of respiration was not homogeneously distributed as assumed by the model, but decreased toward the center of the aggregates. The coefficient of variation of the respiration of 4-mm aggregates placed at 0.21 m(3) O-2 m(-3) was approximately 40%. This interaggregate variability must be incorporated into the model to prevent underestimation of the real and potential rates of respiration. Physical heterogeneities affected O-2 transport within aggregates and induced irregularities in the O-2 distribution. However, there was no evidence of any effect of such heterogeneities on the aggregate respiration.