Heart rate variability (HRV) measures cycle-to-cycle correlations in the instantaneous oscillation period of the heart. In this paper it is shown that a simple model process, consisting of a sum of uncoupled sinusoidal oscillators with slightly different frequencies, has a HRV spectrum with a 1/f scaling over a range of frequencies. This implies that the appearance of 1/f HRV spectra in experiments should not be considered evidence of oscillator coupling or other more complex dynamics. The origin of the 1/f scaling in the model is examined analytically, and its dependence upon the sampling of low-amplitude fluctuations of the process is highlighted.