We randomly placed 10 parkinsonian patients on high- and low-protein diets that tasted and looked alike, each for 1 week. All patients were taking L-dopa and carbidopa with or without other antiparkinson medications; medications remained unchanged. A "blind" physician recorded the modified Columbia scores, objective measurements of rigidity, movement velocity, and pegboard tests three times a day for 5 days during each week. The patients recorded fluctuations hour to hour. We measured serial blood L-dopa levels on day 4 of each week. Performance was significantly better while the patients were on low-protein diets. These results did not correlate with blood L-dopa levels, which had higher peaks in three patients while they were on high-protein diets despite inferior performance and increased number of "off" hours. Thus, high dietary protein probably affects the efficacy of L-dopa at a central level.