The loess region of western Shanxi has typical geomorphological features of the eroded tableland gully. The duration of vegetation restoration in this area has a relatively long history. Exploring the long-term effects of vegetation restoration on soil physical and chemical properties, as well as carrying out long-term field observation of environmental factors, has significant implications and supports for achieving regional ecological restoration and protection. This study selected four typical vegetation restoration forests (i.e. secondary forest of Quercus wutaishanica Mayr, artificial forest of Robinia pseudoacacia L., artificial forest of Pinus tabuliformis Carr., and artificial forest of Platycladus orientalis) in the loess region of western Shanxi as the research objects. The soil physical and chemical properties of different soil layers (0-10 cm, 10-20 cm, 20-40 cm, 40-60 cm, 60-80 cm, 80-100 cm) were measured in 2006, 2012, 2017, including organic carbon, total nitrogen, total phosphorus, total potassium, pH, soil bulk density, sand, silt, clay. The statistical methods such as principal component analysis (PCA), redundant analysis (RDA) and correlation analysis were used to explore interactions among various factors in the soil under long-term vegetation restoration, and analyze the effects of vegetation types on spatiotemporal heterogeneity of soil properties. The results showed that during the long-term vegetation restoration process, soil organic carbon and total nitrogen contents in different forests initially increased then decreased, while soil total potassium and total phosphorus contents decreased at first stage and then increased. Vegetation restoration enhanced the accumulation of soil organic carbon and total nitrogen. Changes in soil pH, bulk density, sand, silt, and clay measurements were not noticeable-these properties were less influenced by external factors during the vegetation recovery period. Overall, long-term vegetation restoration increased the content of soil nutrients, soil sand, and soil silt, while reducing soil bulk density and soil clay. Compared with artificial forests, the secondary Quercus wutaishanica Mayr forest was the best vegetation type as regards of improving soil structure and increasing soil fertility. Redundancy analysis indicated that there existed the differences in primary attributes affecting soil nutrient content under the condition of long-term restoration of different vegetation types. Silt in the soil of Robinia pseudoacacia L. and soil bulk density of Platycladus orientalis made the highest contributions to the changes in soil nutrients over long-term scale. The results of this study can provide data support and a theoretical basis for assessing the ecological benefits of long-term vegetation restoration practices in loess eroded tableland areas.