Rotation is an important cultural practice that farmers use to manage crop growth and diseases. Diversified crop rotations directly affect plant residue inputs that may enrich specific groups of microbes that form association with the following crop. At two field sites, we observed yield differences in soybean planted after the following four long-term (14 year) crop sequences - continuous corn(CC), corn/corn/soy(CCS), corn/soy/corn(CSC), soy/corn/soy(SCS). Soybean yields were in decreasing order, CC>CCS>CSC>SCS, and the yield differences could not be explained by soil chemical and physical properties. Our goal was to relate soil health indicators and soil microbial community composition with the differences in soybean yields. Soil protein, permanganate-oxidizable carbon (POX-C), and beta-glucosidase activity were measured for soil health. After 14 yr of the rotation regimes, soil protein was higher in CC plots compared with other rotations. POX-C was also greater in CC plots. Cyanobacteria were associated with higher POX-C and protein levels. Three taxa of bacteria and one taxa of fungi were associated with higher levels of beta-glucosidase and correlated with the lower soybean yield observed in the SCS treatment. We also found differential abundance of specific bacterial and fungal OTUs to be informative on predicting yield differences, especially fungal pathogens from genera Macrophomina and Corynespora at one location. Additionally, random forest classifier was effective at identifying informative OTUs that were associated with rotational treatments and soybean yields.