Optimal fitting of a myocardial time-activity curve is accomplished with a monoexponential plus a constant, resulting in three parameters: amplitude and half-time of the monoexponential and the constant. The aim of this study was to estimate the precision of the calculated parameters. The variability of the parameter values as a function of the acquisition time was studied in 11 patients with cardiac complaints. Of the three parameters the half-time value varied most strongly with the acquisition time. An acquisition time of 80 min was needed to keep the standard deviation of the half-time value within ±10%. To estimate the standard deviation of the half-time value as a function of the parameter values, of the noise content of the time-activity curve and of the acquisition time, a model experiment was used. In most cases the SD decreased by 50% if the acquisition time was increased from 60 to 90 min. A low amplitude/constant ratio and a high half-time value result in a high SD of the half-time value. Tables are presented to estimate the SD in a particular case.