The actual attempt was finally well below 55 or 56 km/h. He started slower than Obree and needed about 20 km to match up with Graeme Obree's split times. From kilometer 12 to 32 his average speed was around 53 km/h, then he increased his speed and at 36 kilometers he was going for 53.094 km for the hour. But then he slowed down again and finally reached 53.040 km. He used a Pinarello carbon bike with a very aerodynamic frame and Campagnolo disc wheels.

There is a video of the full attempt on youtube. In this video the lap times are not shown, just the times for each kilometer. So one can create a graph similar to that above:

The comparison with Graeme Obree's attempt:

As stated in an article on  "53x12.com" by Dr. Michele Ferrari , it is likely that Indurain, only 5 weeks after the finish of the Tour de France, was not at his peak form when he set his record. He also points out that Indurain's position on the bike didn’t seem to be optimal (however, as I mentioned above, it was good enough to win all the major time trials on the road...)  A key factor could be that the rider’s track speed is not constant, presenting continued accelerations and decelerations corresponding to curves and straight-aways. A heavy rider pays dearly for these accelerations—far more than a light-weight. Finally, at these speeds the one-hour record, Indurain clearly suffered more from the centrifugal force in the curves, since he is about 15 kg heavier than the "normal" cyclist. (Note that Ondrej Sosenka  (2 meters tall!) chose the track in Moscow (for his UCI hour record 2005) which has a length of 333 m,and therefore less centrifugal forces...)

Video of the hour record attempts 1994 (Obree, Indurain, Rominger) on Youtube

Video on Youtube with the full attempt (in Spanish)