In 140 adult patients with FEV1 greater than or equal to 60 percent predicted referred because of suspected asthma, we compared responses to methacholine and isocapnic cold-air hyperventilation. Most challenges were accomplished on the same day, cold air always being done first. The cold-air test employed a single episode of hyperventilation (target, 30 times the FEV1), and subsequent FEV1 changes were noted, a decrease of 10 percent being defined as a positive test result. For methacholine, subjects inhaled aerosols of increasing concentrations and the dose associated with a 20 percent decline in FEV1 (PC20) was noted; positive tests were defined as a 20 percent decrease at concentrations less than or equal to 8 mg/ml. Of the 140 patients, 65 had negative results on both challenges. Twelve patients had positive results on cold-air testing but negative responses to methacholine, and 17 had the opposite result. Among patients with positive results to either test, there was a significant correlation (p less than 0.001) between change in FEV1 with cold air and log PC20, but there was considerable scatter, the results of one test accounting for 25 percent of the variability in the other. Some scatter may have been related to the methods we used, but much was probably due to the patients themselves. Neither test should be used on an exclusive basis to make the diagnosis of asthma.