Prime time versus nonprime time on MEDLINE.

The number of users on the MEDLARS system varied more during nonprime time than on prime time; there were times of higher than average use and times of much lower use. This variation in the number of users influenced search time. The study found a variation of as much as 19% from the mean search time for times of high use versus times of low use. Added-time formulas showed the additional time allowed for a search on nonprime time before costs would equal those for the same search on prime time. (For example, a search with four citations that took five minutes on prime time could take up to eight minutes and thirty seconds on nonprime time before costs would be equal.) In all examples, nonprime time searching was the most cost-effective alternative.