The objective of this research was to examine the conditions and characteristics of soil subgrades that had been stabilized using mechanical compaction. Goals of the study are to identify and examine the engineering properties and behavior of the “soft layer’ of material observed at the top of untreated, highway pavement soil subgrades. Alternative methods of preventing, or mitigating, the development of the soft layer are discussed. Evidence is presented that shows that a soft layer of soil frequently develops at the top of untreated, highway soil subgrades. Data are presented that show strengths obtained from mechanical compaction are largely destroyed when untreated compacted soils are exposed to moisture. CBR values of compacted clayey soils initially are high but become small when exposed to saturation. In situ CBR values measured at the tops of untreated subgrades, where mechanical compaction was the only means used to stabilize the soil subgrade, were smaller than unsoaked and soaked laboratory Kentucky CBR values. At the 85th percentile test value, the laboratory KYCBR value of compacted, unsoaked clayey specimens was 11.5 while the CBR value of soaked specimens was 3.0. For comparison, the in situ CBR value of untreated subgrades at the 85th percentile test value, as shown in this study, was only 2. Using a bearing capacity model, based on limit equilibrium of layered media, bearing capacity analyses of flexible pavement sections were performed. The analyses show that when the in situ CBR is equal to or below 3, the pavement was unstable, i. e., the factor of safety against failure was 1.0 or below. However, when the in situ CBR value was 6, or greater, the pavement was generally stable and the factor of safety was 1.5, or greater. Chemical admixture stabilization of soil subgrades is the most effective means of maintaining large CBR values during construction and throughout the life of the pavement. In situ CBR values at the 85th percentile of tests performed on the tops of soil subgrades treated chemically with lime kiln dust, hydrated lime, and Portland cement and that had been in place for 8 to 15 years were 24, 27, and 59, respectively. At the 85th percentile test value, in situ CBR values of chemically treated subgrades were about 12 to 30 times larger than the in situ CBR value of 2 of untreated subgrades.