We explore the feasibility of using mutual information to characterize the homeopathic effect. This quantity measures the information gained about a signal at time (t + tau), from its value at an earlier time t; it quantifies the predictability of data. We illustrate our method with an analysis of the homeopathic effect of Strophanthus hispidus on the cardiac rhythm of healthy human subjects, using data from a previous experiment. Our results allow an intuitively clear rendering and agree with the similitude principle applied to this case. They also show that the solvent has a significant effect on the signal; hence, it does not act as an ideal placebo and we discuss some therapeutic corollaries to this observation.