A new method using the covariance function as a measure of functional similarity is presented for dynamic analysis of a sequence of scintigraphic cardiac images taken throughout the cardiac cycle. The similarity between the temporal response of pixels in a reference region of the scintigraphic image series and the temporal response of the remaining pixels in the image sequence is calculated. The resulting covariance image is a functional image representing regions with different temporal dynamics. A box-plot representation of this image permits better interpretation for clinical decision making. This analysis allows visualization of the ventricular emptying pattern, which may be useful in studying motion or conduction abnormalities. Compared to Fourier analysis, our method does not make assumption that the data are periodic and that the transition between the first and the last frame of the study is smooth. The proposed method has been performed in one normal patient and twenty patients with abnormal ventricular contraction patterns, and there is no computational difficulty in its implementation. A comparison with the Fourier analysis is performed.