A special test station was designed and developed to examine self-lubricating composite wear rate and the mechanism of transfer film formation that occur in bearings operating with composite retainers. The apparatus is constructed to approximate (a) the sliding contact between the ball separator pockets and the balls and (b) the rolling contact between bearing balls and races. The ball bearing simulator is capable of duplicating varying ranges of (a) ball/race rolling velocities (unidirectional or oscillating); (b) ball/race (Hertzian) bearing loads; (c) ball/composite loads; and (d) total included ball/race oscillatory angles. The experimental arrangement provides for in situ photomicrography of the ball and race paths and interferometric thickness determination of sufficiently transparent lubricative transfer films (e.g., formed by reinforced PTFE/MoS2composites). This test station can be used as an intermediate tool for understanding the quantitative effects of bearing (operational) parameters on composite retainer wear and film formation. A method is proposed whereby excessive wear of the retainer is designated to be the primary failure mode of self-lubricated ball bearings. Composite wear, however, can be quantitatively described by factorial (empirical) wear equations. These equations are derived from computer regression analysis of data obtained from statistical test matrices of bearing simulator tests. Simultaneous determination of transfer film thickness and ball/race damage can eventually lead to a unified, empirical method of self-lubricated bearing performance prediction.Presented as an American Society of Lubrication Engineers paper at the ASLE/ASME Lubrication Conference held in Miami Beach, Florida, October 21–23, 1975