Abstract: Accurately dating the age of basalt has been an important scientific issue for geologists. Due to its unique texture and composition, the basaltic minerals are limited to certain types, with scarce zircons in tiny grain sizes. It is difficult to pick pure minerals such as zircon from basalt in physical ways, therefore, studying the formation age of basalt by internal isochron or zircon U-Pb methods is challenging. Since whole rock basalt samples, on the other hand, have a limited range of ¹⁴⁷Sm/¹⁴⁴Nd ratios, low Rb content and low Rb/Sr ratio, the whole rock Sm-Nd method or Rb-Sr method usually cannot yield an accurate age. In this study, the principle of internal isochron and chemical methods, using aqua regia and Hf-HNO₃ to dissolve the same basalt samples by steps is utilized, and performed by Sm-Nd isotopic analysis on the 3 phases from the same sample: 1) dissolved phase in aqua regia, 2) undissolved phase in aqua regia and 3) whole rock phase. The results shows that, through dissolving the samples in different acid media by steps, quartz-diopside-feldspar mineral assemblage can be extracted from basalt, and this assemblage has the same ε_Nd(t) and Nd modal age as the whole rock. In the internal isochron constructed by minerals and whole rock, the variation range of ¹⁴⁷Sm/¹⁴⁴Nd ratio increases from 0.005 within the whole rock to 0.11, the ¹⁴³Nd/¹⁴⁴Nd ratio range varies from 0.512500-0.512547 within the whole rock to 0.512500-0.513145. The Sm-Nd age yielded with this method is identical to previous zircon U-Pb age within uncertainty: t=(991±21) Ma, MSWD=2.1. By comparing the results from conditional experiments, we propose that Sm-Nd internal isochron dating by phase separation is more suitable for rock samples of Precambrian or older. The establishment of this method can effectively increase the success of Sm-Nd isochron dating for basalt, as well as offer a new estimate for other aphanitic rock samples that cannot provide pure minerals for easy dating.