Helmet testing has been carried out using a dummy with a flexible neck, to identify the events that occur in impacts occur to the side of the head. The levels of angular acceleration correlate with the peak linear accelerations; the peak values occur before the large angular motions of the head. For the better designs of helmet, tested at the side with an impact energy equivalent to 30 J in a rigid anvil test, neither value is considered to be injurious. The load spreading efficiency of the rigid foams used in soft shell bicycle helmets is good so long as the foam does not bottom-out at the impact site, and there is evidence that polypropylene foam is better at load spreading than polystyrene foam. We conclude that the side of the head can be protected by a suitable helmet design.