One of the most important considerations when irrigating with marginal water or managing sodic soils is maintaining or restoring good soil structure. Recommendations developed to date have considered water quality parameters and assumed that soils behaved similarly. We examined the flocculation behavior of three different types of micaceous soils at variable electrolyte concentration, sodium adsorption ratio [SAR =Na/(Ca + Mg)0.5, where concentrations are expressed in mmol L−1], and pH. The variability in critical coagulation concentration (CCC) among soil types and among samples within the same soil type also was investigated. The CCC values were calculated at electrolyte concentrations of 2, 5, 10, 20, and 40 mmolc L−1 at SAR values of 5, 10, 20, 40, and 80 (mmol L−1)0.5, and at pH values of 6.0, 7.5, and 9.0. The variability in CCC was large among soil types of similar mineralogy, and even within the same soil type. This variability was greater at higher pH, suggesting that at least part of the variation results from differences in variable charge. Increasing pH had an adverse effect on dispersion, especially when the SAR was high enough to break apart the illite domains. There was no relationship among CCC, organic matter, and Fe and Al oxide contents for the soils studied. In the range investigated (1–4 g organic C kg−1 soil), only the Ebro Basin soils showed increases in CCC with increasing organic matter at high pH and SAR. Aggregate stability was measured on aggregates of the soils after saturation with solutions having salt concentrations of 2 and 40 mmolc L−1; SAR values of 5, 20, and 80 (mmol L−1)0.5; and pH values of 6.0, 7.5, and 9.0. No relation was found between the flocculation test data and the aggregate stability results. Apparently the measured parameters in these two tests are governed by different soil factors. Particle-size distribution and porosity, as well as the drying and rewetting processes, exert a major influence on aggregate stability, but are not considered in flocculation tests. Regardless of the test used, differences in results for samples of the same soil type preclude the use of generalized threshold lines for soil stability.