The origin of the "S3" EPR signal from calcium-depleted photosystem 2 samples has been investigated. This signal is observed after freezing samples under illumination and has been assigned to an interaction between the manganese cluster and an oxidized histidine radical [Boussac et al. (1990) Nature 347; 303-306]. In calcium-depleted samples prepared by three different methods, we observed the trapping of the tyrosine radical YZ+ under conditions which also formed the "S3" signal. An "S3"-type signal and YZ+ were also formed in PS2 samples treated with the water analogue ammonia. Following illumination at 277 K, the "S3" and YZ+ signals decayed at the same rate at 273 K in the dark. Both the YZ+ and "S3" signals decayed on storage at 77 K and could be subsequently regenerated by illumination at 8-77 K. No evidence to support histidine oxidation was found. The effects of DCMU, chelators, and alkaline pH on the dark-stable multiline S2 and the "S3" signals from calcium-depleted samples were determined. Both signals required the presence of EGTA or citrate for maximum yield. The addition of DCMU caused a reduction in the yield of "S3" generated by freezing under illumination. Incubation at pH 7.5 resulted in the loss of both signals. We propose that a variety of treatments which affect calcium and chloride binding cause a stabilization of the S2 state and slow the reduction of YZ+. This allows the trapping of YZ+, the interaction with the manganese cluster (probably in the S2 state) resulting in the "S3" signal. The data allow the position of the manganese cluster to be estimated as within 10 A of tyrosine Z (D1-161).