This study shows how agricultural management practices by indigenous people as well as soil attributes relate to the diversity and composition of soil bacterial and fungal communities in hill rice ecosystems prevalent in the northeastern region of India. Soil samples were collected in three different rice ecosystems viz. direct-seeded upland slope, transplanted upland terrace and low land from two different locations in Ri-Bhoi district of Meghalaya. An array of soil attributes {moisture content (MC), soil particle distribution (sand, silt and clay contents), pH, total organic carbon (TOC), total nitrogen (TN) and cation exchange capacity (CEC)} were determined. The diversity and composition of bacterial and fungal communities were explored by denaturing gradient gel electrophoresis' (DGGE) and ribosomal internal-transcribed spacer analysis (RISA) fingerprinting, respectively and community data were analyzed using nonmetric multidimensional scaling ordination (NMDS). Relationships between soils attributes and community data were deciphered by principal component analysis (PCA). Bacterial phylotype richness (Margalefs index) and diversity (Shannon's index) were higher in soils of lowland rice ecosystem compared to that in other rice ecosystems; whereas these indices for fungal community found highcr in disturbed rice ecosystems (slope land and young upland terrace). The NMDS plot indicated that community composition varied significantly betwecn rice ecosystems. Soil MC and management practices in rice ecosystems seem to be the strongest determinant of bacterial and fungal community composition and thesc two determinants strongly influenced TOC, TN, sand and silt content, and CEC and hence, these soil attributes also found to be co-determinants of bacterial and fungal community composition in rice soils. Soil pH was found to be a week determinant of community composition in rice ecosystems. Overall, results suggested that the structmes of soil bacterial and fungal communities are not random at ecosystem scale and that diversity and comumnity composition of bacteria and fungi in rice ecosystems can largely be predicted with management practices, soil MC and the covariant of soil MC like TOC, TN, soil texture and CEC.