The use of likelihood cross-validation for guiding the scale development process is formulated and demonstrated, including choosing the number of factors, assessing item-factor allocations suggested by rotations, adjusting those allocations, reducing the number of factors, removing items, and assessing the applicability of scales to subjects other than those for whom it was originally developed. An example analysis is presented on the development of scales to measure how parents caring for a child with a chronic condition view their family's management of that condition.