Objectives(1) To examine the validity of comparing the phase of broad-band signals. (2) To measure phase synchrony over the whole head, at a variety of frequencies. Methods The concept of broad band phase is investigated (a) by visual comparison of the time series of two channels of filtered data with the time series of the spatial analytic phase difference (SAPD) between the two channels and (b) using artificial sinusoids. Phase synchrony is then measured in 64-channel EEG recorded while human subjects performed a perceptual-cognitive task, by calculation of analytic phase differences between each channel and a frontal synchrony reference channel. The number of channels in synchrony with the reference channel at a series of frequency passbands is compared for data acquired using a common recording reference, the same data re-referenced to an average reference and artificial noise. Results Analytic phase is shown to represent the resultant of the phasor angles of all the narrow band signals incorporated in a composite waveform. Episodic global phase synchrony is identified in background EEG, in all passbands from theta to epsilon. Many of the episodes of widespread synchrony occur in both common-referenced and average-referenced data, but some common-reference episodes are not seen in average-referenced data. In both forms of data, synchrony is about equally widespread in all subjects at lower passbands, but more widespread in some subjects than others at higher passbands. Conclusions (1) It is valid to measure the analytic phase of broad band EEG signals. (2) Non-local phase synchrony is intermittently present in all frequency bands from theta to epsilon, not only during and after external stimuli, but also in background EEG. (3) In some subjects synchrony is more widespread in gamma and epsilon bands than in beta, alpha or delta bands, but in others the reverse is true. (4) Some of the episodes of synchrony seen in common referenced data may be artifacts of a sudden decrease in power at the recording electrodes in comparison with the common reference electrode. However, most of the episodes of synchrony in common-referenced data cannot be explained in this fashion. (5) Episodes of widespread synchrony are not established instantaneously. During the establishment of most episodes of ‘40 Hz’ synchrony, the number of channels in synchrony peaks after about 100 ms. Significance If long-range phase synchrony really is a hallmark of consciousness, it should be present most of the time the subject is conscious. Our results confirm this prediction, and suggest that consciousness may involve not only gamma frequencies, but the whole range from theta to epsilon. The mechanism of synchrony establishment at the scalp as shown by the present method is relatively slow and thus more likely to involve chemical synapses than gap junctions, electric fields or quantum non-locality.