The Timers of the CAB are very accurate and reliable.  Stating that they are accurate to + 1ms is a conservative estimate.  In all cases analyzed to this point there is no variability at the millisecond level.  The variability is found at the microsecond level and when rounded to the millisecond it becomes exactly what is expected. 

Timer accuracy is a crucial part of any tool that involves reaction time and information processing speed data.  Inaccuracies, poor sampling rates (granularity of timers) and built-in variability all limit the tests ability to effectively perform the assessment task consistently.  Poor sampling rates of timers (such as some of those utilized in many DOS based programs) also lose sensitivity by artificially increasing variance.  The timers used in the CAB are extremely accurate and reliable.  This includes all of the operating systems used for the CAB (Windows 95, Windows 98, Windows ME, Windows NT, Windows 2000, and Windows XP).  As long as no other programs are operating at the same time as the CAB.   Statistical analysis have been included as examples of the CAB timer's accuracy.   This data has been reviewed for a number of different computer systems.  The data presented here are for two systems using Windows 95.  A 100 HZ Pentium with 16 Megs of EDO Ram and 256K Cache.  The other system is an AMD K6 200 HZ MMX system with 32 Megs of SDRAM and 512K Cache.  If more information is needed or you would like copies of analyses performed on other platforms please e-mail us at cab@neuropsychworks.com. These results only improve with more advanced computers using Windows XP.  These older computers with older operating systems are used as examples of minimum requirements and their accuracy. 

There is also a copy of Appendix A of the CAB manual which provides information about how the CAB should be run and the built-in tools to assess timer accuracy on any system.