The reports dealing with the effects of weak electromagnetic fields (EMFs) on brain electrical activity have been inconsistent. We suspected that the use of linear models and their associated methods accounted for some of the variability, and we explored the issue by using a novel approach to study the effects of EMFs on the electroencephalogram (EEG) from rabbits and humans. The EEG was embedded in phase space and local recurrence plots were calculated and quantified to permit comparisons between exposed and control epochs from individual subjects. Statistically significant alterations in brain activity were observed in each subject when exposed to weak EMFs, as assessed using each of two recurrence-plot quantifiers. Each result was replicated; a sham exposure control procedure ruled out the possibility that the effect of the field was a product of the method of analysis. No differences were found between exposed and control epochs in any animal when the experiment was repeated after the rabbits had been killed, indicating that a putative interaction between the field and the EEG electrodes could not account for the observed effects. We conclude that EMF transduction resulting in changes in brain electrical activity could be demonstrated consistently using methods derived from nonlinear dynamical systems theory.