Background: Well-differentiated liposarcoma (WDLPS) and dedifferentiated liposarcoma (DDLPS) are common malignancies of adipocytic origin, which are in different differentiation stages. Both of them are characterized by high local recurrence rate. The tumor microenvironment (TME) is associated with the prognosis of tumor and immune-related cell infiltrations has been proved to be one of the mechanisms affecting the efficacy of immunotherapy. However, the difference between TME characteristics of WDLPS and DDLPS as well as the association between immune cell infiltrations and recurrence of WD/DDLPS has not been fully studied. Methods: 37 Chinese patients with liposarcoma were enrolled in this study, including 24 patients with WDLPS and 13 patients with DDLPS. Tissue samples from these patients were examined, who experienced previous recurrence or no recurrence. The density and percentage of the immune-related cell subsets in the TME were identified by multiplexed immunohistochemistry and multispectral imaging. Twenty cell lines derived from patients with liposarcoma in our center were used to analyze the correlation of immune-related cells content between patient tissues and cell lines. Results: Overall, DDLPS had more immune cell infiltrations than WDLPS. Compare to WDLPS, the density of CD8+ (the cytotoxic T cells identification) cells and CD163+ (M2 macrophages identification) cells and the percentage of FOXP3+ (Treg cells identification) cells was significantly more in DDLPS, with a Wilcoxon p-value of 0.02, 0.04 and 0.03, respectively. There was no significant difference in other protein markers between WDLPS and DDLPS. According to the multiple cut-off values (tertiles, quintile and tenths) of immune-related cells, the Chi-square test results equally suggested more immune infiltrations in patients with DDLPS. PD-1+ cells was significantly more (Wilcoxon p = 0.02), while FOXP3+ cells were significantly fewer (Wilcoxon p = 0.04) in patients without recurrence than those in patients with previous recurrence. Patients without recurrence had more CD3+ cells (the total T cells identification) and CD163+ cells infiltrations, although the both Wilcoxon p-value were not statistically significant. The total T cells, CD4+ T cells and Treg cells were more distributed in tumor tissues with no recurrence or recurrence for once, but less in patients with two or more recurrence. In addition, no strong correlation of immune-related cells content between patient tissues and cell lines were observed in our analysis. Conclusions: Our findings suggested that patients with DDLPS might benefit more from immunotherapy than patients with WDLPS due to the infiltrating TME of DDLPS. The increase of recurrence times might lead to a more repressive TME in liposarcoma.