Blood vessels in tumors often exhibit abnormal morphology and function, which promotes the growth, metastasis and resistance of tumors to conventional therapies. Therefore, vascular normalization is an emerging strategy to enhance the effectiveness of radiotherapy and chemotherapy when used in combination; however, there is a lack of evidence regarding the optimal schedule for the co-administration of anti-angiogenic and chemotherapeutic drugs. Scheduling treatment is important as the period for normalization is transient, also known as the 'time window'; however, no biomarker has been identified to detect this window. In the present study, recombinant human endostatin (rhES) was employed as an anti-angiogenic agent in xenograft tumor tissue in mice. Following rhES or control (saline) treatment, the density and integrity of tumor vessels were detected by immunofluorescence staining for cluster of differentiation 31 and α-smooth muscle actin; the level of hypoxia in tumor tissue was examined by immunohistochemistry with pimonidazole; the necrotic area was evaluated by hematoxylin and eosin staining; and the level of thrombospondin-1 (TSP-1) in plasma was tested by ELISA. The Cell Counting Kit-8 assay was also used to evaluate the effect of rhES on the proliferation of colon carcinoma SW620 cells. A 'time window' normalized vasculature was determined between day 4 and 6 following rhES treatment, and accompanied by a decrease in hypoxia in tumor tissue. Decreasing plasma TSP-1 levels were consistent with changes in vascular morphology and hypoxia, which exhibited features of normalization. In addition, rhES had no effect on the proliferation of SW620 cells, suggesting that the reduction in TSP-1 was associated with increased oxygen content during vascular normalization, rather than inhibited cell proliferation. In conclusion, TSP-1 may be a potential biomarker for predicting the normalization window of colon cancer vessels.