To offer guidance for using Brazilian tensile strength (BTS) to estimate UCS of anisotropic rocks, this study uses sandstone (fine or coarse grain) and gneiss (0°, 45°, 90° inclined anisotropy) to investigate the influence of grain size or anisotropy on the correlations of UCS-BTS. According to the regression analysis, there is no significant equation of UCS-BTS for rocks with vertical anisotropy. The grain size variation or multidirectional anisotropy can result in a decrease in the determination coefficient value of correlations. Then, coarse grain size or vertical anisotropy deteriorates the statistical performance of correlations between UCS and BTS, reflected by the Akaike Information Criterion and performance index. For rocks with fine grain size or 45° inclined anisotropy, the data points of estimated UCS are clustered uniformly around the exact estimation line. Finally, the accuracy of predicted UCS via BTS declines obviously following the varying grain size or different anisotropy orientations. Using empirical formulas with different grain sizes or anisotropy properties can generate significant errors in estimated UCS. To predict UCS, BTS should be extracted from rocks with single grain size magnitude or unidirectional anisotropy. Moreover, the Brazilian test parallel to the anisotropy cannot be used to derive the correlation of UCS-BTS.