CrystalDiskInfo reads the health data that hard drives and solid-state drives expose through S.M.A.R.T. and related storage interfaces. It identifies a physical drive, reports temperature and operating history, then lists the raw attributes supplied by that device. CrystalDiskInfo does not scan files, repair sectors or benchmark speed. Its job is to make the drive’s own counters visible before a storage problem becomes an unreadable volume.
+
One drive at a time
+The top area separates detected physical drives. Selecting one changes every detail below it, so a computer with several disks needs careful model and capacity checks before anyone interprets a warning. Drive letters help, but one physical disk can hold several letters and a disk without a mounted volume can still appear. The displayed interface and transfer mode also distinguish a SATA device from an NVMe device or an external bridge.
+CrystalDiskInfo shows power-on hours, power-on count, firmware, serial number, temperature and supported commands when the drive reports them. A privacy option can mask the serial number when copying a report or taking a screenshot. Missing fields do not automatically mean a failed disk; some controllers simply do not expose every value.
+
Health needs context
+The large status label condenses many attributes into Good, Caution, Bad or Unknown. For many hard drives, reallocated sectors, current pending sectors and uncorrectable sectors matter more than a single generic percentage. A pending sector may disappear after a successful rewrite or become reallocated, so its trend and the condition of the stored data matter. For NVMe drives, the manufacturer’s percentage-used and critical-warning values follow a different model.
+CrystalDiskInfo can graph attributes over time when it has collected earlier readings. A rising count is usually more informative than comparing raw hexadecimal values from unrelated models. Manufacturers also define some attributes differently, which means a large raw number is not automatically dangerous. The program’s color is a prompt to inspect and back up, not a guarantee about remaining life.
+Temperature has its own alert threshold. The color can change when the measured value crosses that setting even while the general health remains Good. A sustained high temperature calls for checking airflow and workload, but a brief spike does not identify a damaged file.
+
USB can hide data
+External enclosures sit between CrystalDiskInfo and the drive. Some USB bridges pass S.M.A.R.T. commands, while others return no data or expose only part of it. The official advanced settings contain several USB command types because one translation method does not fit every bridge. Changing that option can make an absent device appear, but it can also stop another enclosure from responding.
+RAID controllers create a similar boundary. A logical RAID volume may be visible to Windows while the controller hides the individual disks. Intel, AMD and other controller paths need compatible drivers and supported pass-through methods. CrystalDiskInfo cannot infer a healthy member state from a logical volume that withholds the underlying counters.
+
Alerts need residency
+Resident mode keeps CrystalDiskInfo running so it can watch temperatures and health changes. Closing the process ends that monitoring. Startup delay and automatic disk detection settings matter on systems whose drives wake slowly, and email notification requires its own configuration. An alert that never fires may therefore reflect a closed process, an undetected external drive or a notification setup problem rather than stable hardware.
+When CrystalDiskInfo reports Caution or Bad, the safest first action is to copy irreplaceable data to separate storage. Tests and repeated reads place more work on the drive. A health utility can document the warning, but it cannot make the only copy of a file safer.
