INTRODUCTION Carbon dioxide (CO2) is considered a potential substitute for CFC/HCFC refrigerants owing to its environmentally benign feature. Due to the low critical temperature of CO2@ the refrigeration system with CO2 as working fluid has to be operated in transcritical cycle under typical ambient conditions for refrigeration and air-conditioning applications. The large entropy generation during the throttling process@ which accounts for 30% to 40% of the total entropy generation@ is thought to be the major reason for comparatively low energy efficiency of the transcritical CO2 cycle (Lorentzen and Pettersen 1993). Replacing the throttling valve with a work-recovering expander in the CO2 refrigeration system is considered to have a great improvement on the coefficient of performance (COP) of the cycle. For the working conditions in which the evaporating temperature is 0??and the heat rejection temperature is 40?? an expander with an efficiency of 70% can improve the COP by 40% to 60% if the isentropic efficiency of compressor is 80% (Nickl et al. 2005). Several kinds of CO2 expanders have been investigated in the last decade. Huff and Radermacher (2003) modified a scroll-refrigeration compressor into an expander to replace the throttling valve in the CO2 air-conditioning system@ and the isentropic efficiency was reported to increase from 30% to 42% after the design improvement associated with the internal leakage in the expander prototype was made. Research work with the similar expander type was conducted by Fukuta et al. (2006) in which the leakage clearance was decreased to 0.01 mm (0.0004 in.)@ and the isentropic efficiency was predicted to reach 55%. Baek et al. (2005a@ 2005b) investigated a piston-cylinder expander@ and the isentropic efficiency was about 11%@ which brought about an improvement in the system COP of 10.5%. According to their studies@ the low efficiency of the expander resulted from the internal leakage. A novel free-piston expander-compressor unit was developed after 1994 by Nickl et al. (2003). It was reputed to have an isentropic efficiency of 65% to 70% for the expander and 90% for the compressor (Nickl et al. 2005). Zhang et al. (2007) proposed another type of free-piston expander-compressor unit@ and the indicated efficiency of the expander obtained from the recorded P-v diagram was 62%. Zha et al. (2003) developed a rolling-piston expander@ and a maximum isentropic efficiency of 50% was reported. Guan et al. (2006) investigated the swing expander@ and the isentropic efficiency ranged from 28% to 44% according to their studies. Recently@ a new expander concept-i.e.@ the two-cylinder rolling-piston expander- was proposed by Yang et al. (2006a) and is being tested in the CO2 refrigeration system. Among the potential expander concepts for transcritical CO2 cycle@ the rotary vane has a simple structure@ is easy to manufacture@ and has no need for inlet/outlet valves. The rotary vane expander was developed and used in a low-grade energy organic Rankine cycle by Badr et al. (1984@ 1985a-e)@ and several issues associated with the expander@ such as the charging processes@ vane dynamics@ and internal leakage were investigated extensively. For the transcritical CO2 refrigeration applications@ Fukuta et al. (2003) investigated a rotary vane expander prototype modified from an oil pump both theoretically and experimentally. According to their studies@ the calculated volumetric efficiency of the expander changed from 0.2 to 0.7@ and the isentropic efficiency changed from 0.2 to 0.4@ depending on the rotational speed. The experimental results of the volumetric and isentropic efficiencies were reported to be 0.64 and 0.43@ respectively@ and the internal leakage was reported to be the main reason for low performance. Edwards (1972) investigated the friction loss in the rotary vane compressor-expander unit in an air-refrigeration system and claimed that the friction loss was about three times larger than the expander work output. In the authors' previous work (Yang et al. 2006a@ 2008a@ 2008b)@ serious leakage and friction losses within the expander were found to be responsible for the low volumetric efficiency and isentropic efficiency and@ therefore@ the investigation of leakage and friction distributions within the expander are the major objective of the current study. First@ a leakage bench test was made to identify the main leakage paths@ and the most serious leakage source due to the loss of contact between the vanes and the cylinder wall was investigated by means of a P-??diagram and the vane movement visual analysis. Then@ the friction distribution was experimentally investigated with the friction bench test rig@ and the major influencing factors were evaluated quantitatively. Based on the experimental results of the leakage and friction distributions@ corresponding design improvements were proposed and validated.