We applied a simplified version of the method suggested by Sugihara-May (Nature 1990: 344: 734-41) to study the control of heart rate (HR) in subjects with diabetes mellitus. The method aims to predict the future of an observation, if a series of observations on the same phenomenon is available. The method quantifies the fact that the series is predictable more or less longtime in the future. A random series is only shortly predictable in the future. HR and blood pressure were measured from beat to beat (by the Finapres system) for about 0.5 hours in 11 subjects with diabetes mellitus and normal blood pressure (group D) and in 10 controls subjects (group N). The subjects were sitting in a temperature-controlled quiet room, isolated from all external stimuli. The 2 groups were matched for age, and had the same weight and height. No difference was observed in mean-value and standard deviation (SD) of BP and HR between the 2 groups. Groups N/D: SBP = 112 +/- 11/123 +/- 11 mmHg, NS; DBP = 64 +/- 9/67 +/- 12 mmHg, NS; HR = 70 +/- 10/69 +/- 7 b/min, NS. Standard deviation of PAS = 5.5 +/- 1.6/5.7 +/- 1.9 mmHg, NS and SD of DBP = 3.5 +/- 0.9/3.4 +/- 1.2 mmHg, NS. The SD of HR (3.0 +/- 0.5/2.3 +/- 1.0 b/min in groups N/D) was somewhat lower in diabetics than in control subjects but the difference was not significant.(ABSTRACT TRUNCATED AT 250 WORDS)