All measurements of EHD film thicknesses have been carried out in simulated test machines. This study uses an actual bearing. A test rig using a 65 mm bore radial cylindrical roller bearing has been constructed with a specially designed sapphire window in the outer track. Full loads, and speeds to 3000 rpm were applied. With specially polished rollers and chromic oxide coating on the window excellent interferometric film thickness measurements were found possible. A Xenon flash lamp was used and a Xenon laser of 0–50 pps, pulse half width of 150 ns and peak power of 100 watts was developed for this research. A microscope and 35 mm camera as well as video tape were used for recording results. Arrangements were made to study any chosen roller and the side of the bearing was also open to view. First the film measurements, when corrected for inlet zone viscous heating, agreed admirably with theoretical predictions for mid and exit film thickness. The effect of inlet boundary length on the film was then investigated in some depth. Studying the effect of the multiple roller system, a number of techniques were used to demonstrate that the inlet boundary length, which controls the lubricant film thickness, was itself controlled by the film thickness between the rollers and track in the unloaded zone. The ribs of oil, formed at either edge of the roller, are only secondary sources of oil for replenishment of the inlet film. It is in fact usual (as shown by the convex shape of the inlet zone) for oil to feed out of the inlet zone into the ribs. Oil globules were sometimes observed riding on an air cushion at the entry to the roller-track conjunction, though completely inoperative as providers of oil.