1. 1. A total of 245 ballistocardiographic smoking tests were done on healthy male medical students, all of them smokers. Sixty-nine subjects (Series O) had one test each at a random time in the day. To test the reproducibility of results 32 men had four tests each, two before breakfast and two before lunch (Series F) and 6 subjects had eight tests each under circumstances kept as standard as possible in every way (Series E). 2. 2. The findings on a single smoking test in 113 students previously described, were essentially unchanged when the nonsmokers and female students were excluded, leaving the 69 subjects of Series O. 3. 3. In 32 male smokers (Series F), smoking a cigarette before lunch produced results similar to those in Series O: a highly significant mean rise in systolic and diastolic pressure, heart rate, and cardiac output, and a fall in stroke volume. Pulse pressure was little affected. Compared with these tests, those performed before breakfast showed lower control readings, with a complementary increase in the response to smoking, so that the final figures after smoking were remarkably constant. Results were essentially the same on a second day. Heavier smokers tended to have lower control values and a bigger response to smoking. Exercise habits influenced the results little. 4. 4. In the eight tests repeated on each of 6 subjects, the first test appeared to be an unbiased estimate of the mean. The reproducibility of the results was calculated, and it proved to be an individual characteristic, that is, mean confidence limits are of value in groups of subjects only. Analysis of variance shows that the change in systolic and diastolic pressures, heart rate, and cardiac output after smoking are likely to prove of value in classifying the subjects, but that for the other two measurements, the “overlap” of individual variation appears to be too great. 5. 5. Comparison of the three series of tests revealed no great differences. The results in Series O, Series E, and the tests before lunch in Series F were substantially the same. The findings in Series F before breakfast were the only exceptions: in the main, control readings were lower and the response to smoking proportionately greater at that time than in any other series of tests. However, this enhancement of response seems of little practical value. 6. 6. From these findings it appears that the ballistocardiographic smoking test, in which blood pressure, heart rate, stroke volume, and cardiac output are measured, gives results which are sufficiently reproducible for use as a screening test to classify young adults according to their patterns of circulatory reactivity. The results were materially unchanged by attempts to standardize further the test conditions, except that on the average, the response was more marked in the fasting state.