The design of a CUSUM control chart typically involves choosing the su
bgroup size and sampling interval to achieve specified control chart p
erformance. One can maintain a specified sampling intensity by using s
mall subgroups at frequent intervals or larger subgroups at infrequent
intervals. We use the average time to signal to investigate the perfo
rmance of a CUSUM chart under alternative selections of the subgroup s
ize and sampling interval. Recommendation are provided for both initia
l-state and steady-state performance.