Pediatrics. 1978 Jun;61(6):811-7.

Relative effects of drugs and diet on hyperactive behaviors: an experimental study.

Williams JI, Cram DM, Tausig FT, Webster E.


In a test of Feingold's hypothesis that food additives trigger the hyperactive response, 26 hyperactive children were randomly assigned to treatment conditions whereby they were given active or placebo medications in combination with challenge cookies with artificial food colors or control cookies without the additives. The children were crossed over into each of the four treatment conditions and experimental procedures were employed, including double-blind assessments through the completion of behavior checklists, by teachers and parents. Stimulant medications were clearly more effective than diet in reducing hyperactive behavior. The parent ratings indicate strong drug effects and inconclusive diet effects. Drug effects are marked in teacher ratings as well. However, when the children were receiving placebos, their hyperactive behaviors in the classroom were greater when eating cookies with artificial colors than when eating cookies without artificial colors. According to the ratings, approximately seven children were no longer hyperactive. There is evidence to suggest that the behavior of three to eight children was diet-responsive, depending on the criteria used. There is evidence, particularly in teacher ratings, in support of Feingold's hypothesis if it is modified. Further research is required to specify which subtypes of hyperactive children respond to a diet free of artificial food colors.

  1. Williams proposed to test the Feingold diet, but he modified it:

    • The diet he used eliminated synthetic food coloring and flavoring.
    • But it did not eliminate either salicylate or preservatives.
    • Both the medication and placebo pills contained the colors eliminated by the diet.

  2. The challenge cookies used had only 13 mg food dye each - a very low dose.

  3. Williams had no "washout" periods between treatment conditions.

  4. Researchers admitted filling in missing data with "estimates."


  1. Parent ratings: Drugs plus diet was better than drugs alone.

  2. Teacher ratings: Food dye challenge without drugs produced worse behavior than placebo challenge without drugs.

  3. 7 out of 26 (27%) of the children were no longer hyperactive, even using this "modified" diet.

  4. CONCLUSION: There is evidence, particularly in teacher ratings, in support of Feingold's hypothesis if it is modified.

Note: If it is modified? You mean one is required to eat salicylates and preservatives to make the diet work? Well, that is slightly weird.

Nevertheless, this study agrees with what parents tell us -- those who do use diet and drugs together find that the combination works better than drugs alone, and with less medication.