Introduction We previously found that mesenchymal stem cells (MSCs) injected intravenously could attenuate peritoneal adhesion by secreting tumor necrosis alpha-stimulating gene (TSG)-6, while MSCs injected intraperitoneally could not. However, the underlying mechanism remains unclear. This study was designed to investigate the means by which MSCs exert their effects. Methods Rat bone marrow-derived MSCs/red fluorescent protein (RFP) were injected either intraperitoneally or intravenously into Sprague-Dawley (SD) rats at different time points after peritoneal scraping. Peritoneal adhesions were evaluated macroscopically at day 14 after scraping. The distribution of MSCs injected intraperitoneally or intravenously was traced by two-photon fluorescence confocal imaging and immunofluorescence microscopy. The co-localization of MSCs and macrophages in the lung and the spleen, and the expression of TSG-6 in MSCs trapped in the lung or the spleen were evaluated by immunofluorescence microscopy. The concentration of TSG-6 in serum was evaluated by ELISA. After intravenous injection of TSG-6- small interfering (si) RNA-MSCs, the expression of TSG-6 in MSCs and the concentration of TSG-6 in serum were reevaluated, and peritoneal adhesions were evaluated macroscopically and histologically. Results MSCs injected intraperitoneally failed to reduce peritoneal adhesion, and MSCs injected intravenously markedly improved peritoneal adhesion. Two-photon fluorescence confocal imaging showed that MSCs injected intravenously accumulated mainly in the lung, where they remained for seven days, and immunofluorescence microscopy showed few MSCs phagocytosed by macrophages. In contrast, large numbers of MSCs accumulated in the spleen with obvious phagocytosis by macrophages even at 4 hours after intraperitoneal injection. Immunofluorescence microscopy showed that MSCs that accumulated in the lung after intravenous injection could express TSG-6 within 12 hours, but TSG-6-siRNA-MSCs or MSCs accumulated in the spleen after intraperitoneal injection did not. ELISA showed that the concentration of TSG-6 in serum was increased at 4 hours after intravenous injection of MSCs, while there was no increase after injection of TSG-6-siRNA-MSCs or after intraperitoneal injection of MSCs. Moreover, intravenous injection of TSG-6-siRNA-MSCs failed to attenuate peritoneal adhesion. Conclusions Our findings suggest that intravenously injected MSCs accumulated in the lung and attenuated peritoneal adhesion by secreting TSG-6, but intraperitoneally injected MSCs were phagocytosed by macrophages in the spleen and failed to attenuate peritoneal adhesion.