In order to accommodate distributed generation (DG) while ensuring the security and efficiency of distribution networks and considering the reasonable interest of transmission system, this paper proposes a spatial-temporal comprehensive matching evaluation, analysis and improvement method for distribution network with DG. The method generally uses a two-stage planning strategy, namely first total then distribution. Firstly, in the first stage, aiming at the total load/DG in a feeder, a comprehensive matching evaluation index system and an evaluation method are proposed, including load and DG matching, net load and feeder capacity matching and transmission and distribution matching. Total matching evaluation indices can evaluate the comprehensive matching level of distribution network with DG from total perspective. Influence factor indices are used to analyze the dominant factors restricting the total matching. In the second stage, under the premise that the total is matched both in size and in time, distribution matching evaluation is further performed in space, namely, considering the location of loads and DGs in the feeder. Secondly, the causes for mismatch are analyzed and improvement measures are proposed for various mismatched scenarios. For total mismatch scenarios, sensitivity analysis is used to find cost-effective improvement factors and guide the improvement of load or DG power curve to achieve better total matching. For total match and distribution mismatch scenario, an optimal placement method is used to adjust the location of loads and DGs in the feeder to achieve spatial-temporal matching. Finally, a case study is presented to demonstrate the proposed method. Six scenarios are studied considering different matching evaluations and different combinations of dominant factors and improving factors. The results show that the method can evaluate the DG-load-grid matching and provide solutions for mismatching scenarios. Through summarizing and comparing various scenarios, proposals for distribution network planning are concluded: total matching should be ensured first, which is reflected in energy matching and smooth net load curve. Then, the spatial-temporal matching can be achieved by optimizing the location of loads and DGs in the feeder.