The relationship between soil heterotrophic respiration (Rh) and soil moisture has been often studied using disturbed soil samples and simple gravimetric and volumetric soil moisture indicators. The objective of this study was to investigate the relationship between Rh and soil moisture in terms of water-filled porosity (WFP), matric potential (Ψm), and relative soil gas diffusivity (Dp/Do) using undisturbed soil cores obtained under different land covers. Soil CO2 efflux, WFP, and Ψm were measured in undisturbed soil samples (250 cm3) collected in the 0–5 cm soil layer (without any vegetation or living roots) under laboratory conditions by combining a CO2 gas analyzer, a scale, and precision mini-tensiometers. For each site and land cover, we also measured soil chemical properties, soil physical properties, and soil microbial composition using phospholipid fatty acid analysis. Grassland soils had the largest total microbial biomass (6275 ng g−1), followed by soils from riparian (5327 ng g−1), and cropland (2745 ng g−1) sites. Bacteria were the dominant group representing 46% (SD = 5%) of the total microbial biomass across all sites and land covers. Maximum Rh was 1.88 (SD = 0.40) μmol CO2 m−2 s−1 in grassland, 1.64 (SD = 0.82) μmol CO2 m−2 s−1 in riparian, and 0.94 (SD = 0.56) μmol CO2 m−2 s−1 in cropland soils. Considering all land cover and soil types, our observations revealed that peak Rh occurred at mean WFP = 0.81 ,Ψm = −6 kPa, and Dp/Do = 0.003. Thus, we recommend avoiding the traditional field capacity definition of −33 kPa for representing peak microbial activity. Water-filled porosity was a more consistent predictor of Rh than Ψm or Dp/Do across soils with contrasting organic matter content, total microbial biomass, soil texture, and soil structure.