ICRU Report 91 recommends two dose gradient metrics for reporting in stereotactic radiotherapy. All of the common gradient indices are one-dimensional although clinical dose gradient problems are at least two-dimensional. Moreover, a loss of accuracy is inherent in two-dimensional dose gradient measures like the superficially averaged dose gradient (SADG) due to the linearization in the third dimension. To close this information gap, the author developed a three-dimensional dose gradient measure called the volumetrically averaged dose gradient (VADG). The correlations between eleven one-dimensional and three two-dimensional dose gradient measures on the VADG were investigated for the linac-based stereotactic radiosurgery of 13 brain metastases. Dose gradient distributions were illustrated by vector fields, dose gradient-frequency histograms, and dose-dependent dose gradient functions. From all of the one-dimensional dose gradient indices, the approximated SADG* showed the strongest correlation on the VADG. The relative errors were in a range of -18.9 to 3.2% and Pearson’s correlation coefficients were r ≥ 0.997. The SADG showed the strongest correlation on the VADG of all of the two-dimensional dose gradient measures with the relative errors in a range of -14.8 to -9.6% and r ≥ 0.999. Three-dimensional dose gradient measures like the VADG best describe dose gradient distributions for lesions that are located in non-homogeneous normal tissue. The quality of each dose gradient measure concerning the description of anisotropic dose gradient distributions is now verifiable. The algorithm for determining the VADG should be implemented in treatment planning and patient plan verification systems to utilize the formalism for all users.