This study investigated the effect of changing the formulas of colicky infants and addressed the methodologic flaws of earlier studies. Attention was paid to issues of designing a blind study, providing a washout period, and measuring and reproducing the effect. In this randomized, double-blind trial, three changes of formula were made: for each of four 4-day periods, colicky infants alternately received a casein hydrolysate formula (Nutramigen) and a formula containing cow milk. Mothers recorded crying in diaries and indicated which crying episodes they considered to have been caused by colic. Nine infants were started on Nutramigen and eight on the cow milk formula. With the first formula change there was significantly less crying and colic in infants when they were fed Nutramigen than when they were fed cow milk (p less than 0.01); with the second change there was less colic when infants were fed Nutramigen (p less than 0.05) but not significantly less crying. By the third change there were no significant differences between formulas. Further analyses demonstrated that there were more clinically meaningful positive responses (a change in crying by at least one third) to Nutramigen than to cow milk (p less than 0.05). However, only one subject had a clinically meaningful response in colic to all three formula changes. These results demonstrate that in some instances, colic improves with elimination of cow milk formula. However, the effect diminishes with time, and only infrequently is the effect reproducible.