Within the dedicated Page statistics group, this counts pages pulled from disk to resolve faults, mirroring the equivalent I/O-group counter but tied to its own gathering schedule. Comparing it to the fault count in the same group tells the admin what fraction of faults in that interval were hard faults requiring actual disk access. For example, a Page-group snapshot showing 200 faults but only 20 reads indicates most faults that period were resolved from lists already in memory rather than from disk.