Objective To explore the imaging findings and diagnostic values of X-ray, CT, MR,and ultrasonography in traumatic knee joints hemarthrosis and lipohemarthrosis. Methods Traumatic knee joints hemarthrosis (12 knees) and lipohemarthrosis (18 knees) proved by operation (27 knees) or puncturation (3 knees) were included in the study. Horizontal-beam plain radiographs (16 knees), CT (30 knees), MRI (30 knees) and ultrasonography (24 knees) in supine position were investigated. Results (1)supine position horizontal-beam plain radiographs: Fat-liquid layer was found in 8 cases of lipohemarthrosis. Dense supragenual bursa was found in 1 case of lipohemarthrosis and 7 cases of hemarthrosis. Fracture (13 knees) was diagnosed correctly. (2) CT findings: double fluid-fluid layer was found in 11 of all 18 cases, and single fluid-fluid layer was found in 7 of 11 cases of lipohemarthrosis. Single fluid-fluid layer was found in 3 of 12 cases of hemarthrosis. Isodensity was detected in 9 cases, and high-density blood clot was found in 4 cases. Fracture (30 knees) was diagnosed correctly. (3) MRI findings: in 12 of 18 cases of lipohemarthrosis, double fluid-fluid layer was shown including supernatant layer as short T1, long T2signal and low signal after fat-suppression, middle layer as long T1, long T2 signal and high signal after fat-suppression, and dependent layer as iso-T1, iso-T2 and slight high signal after fat-suppression. Single fluid-fluid layer was seen in 6 cases, only had aforementioned upper and under layer.Only aforementioned supernatant layer and dependent layer were seen in 12 cases of hemarthrosis. 4 cases showed entire blood clot in fluid, T1WI showed middle signal or center iso-signal accompanied with peripheral high signal ring, and fat-suppression imaging showed high signal. T2WI and fat-suppressionimaging showed middling or high signal accompanied with peripheral low signal ring. Fracture (30 knees) was diagnosed correctly. (4) Ultrasound findings: In 10 of 14 cases of lipohemarthrosis, double fluid-fluid level was shown, supernatant layer as equal echo, middle layer as echoless, and dependent layer as cloudy echo. Four cases with single fluid-fluid level only showed aforementioned upper and under layer. Three of 10 cases of hemarthrosis showed single fluid-fluid level, only showing aforementioned upper and under layer,and 7 cases showed cloudy echo and float. In 3 cases the fluid blood clot showed irregular shape low-equal echo bolus. No fracture hne was found. Conclusions CT can clearly detect fracture line, hemarthrosis and lipohemarthresis, and can substitute plain radiography. MRI is the best way to diagnose hemarthresis and lipohemarthrosis. Ultrasonography can be used in diagnosing hemarthresis and lipohemarthrosis but not helpful in the diagnosis of fracture.