Treatment in heart failure could be guided by additional non-clinical measures, such as neurohumoral levels. Variability in heart rate is known to reflect neurohumoral stimulation. With this in mind, we sought to assess retrospectively the variability in heart rate to guide the treatment of infants in heart failure. We analysed retrospectively the data from 20 infants with a significant left-to-right shunt. All were unsuitable for cardiac surgery or interventional therapy at the time the treatment had commenced. None of the infants improved while receiving diuretics, spironolactone, and digoxin alone, but improved after the addition of propanolol or metoprolol. None of the infants had problems during or after the subsequent operation. Parasympathetic activity reflected by parameters of variability in heart rate, such as the square root of adjacent RR-intervals, and the amount of adjacent RR-intervals greater than 50 milliseconds, improved in nearly all infants during beta blockade. On the other hand, parameters of variability in heart rate reflecting sympathetic activity did not change. Parasympathetic activity reflected the clinical state of nearly all the infants. These parameters, therefore, seem to be a good non-clinical parameter, showing the optimal treatment for heart failure in an ambulatory setting.