The latency of Micrococcus lysodeikticus membrane-bound Mg(2+)-adenosine triphosphatase (ATPase) is expressed by the ratio of its activity assayed in the presence of trypsin ("total") versus the activity assayed in absence of the protease ("basal"). By isolating membranes in the presence of variable concentrations of Mg(2+) (50 mM, 10 mM, or none) and by washing them with different Mg(2+)- and ethylenediaminetetraacetic acid-containing tris(hydroxymethyl)aminomethane-hydrochloride buffers (pH 7.5), we showed that the enzyme latency was dependent on the environmental concentration of this divalent metal ion. Mg(2+) bound to at least two classes of sites. The binding of Mg(2+) to low-affinity sites (saturation at approximately 40 mM external Mg(2+)) induced a high basal ATPase activity, whereas its binding to medium-affinity sites (saturation at about 2 mM Mg(2+)) correlated with low basal activity and a very high stimulation by trypsin. Membranes with tightly bound Mg(2+) (high affinity?) revealed an intermediate behavior for the latency of M. lysodeikticus ATPase. The Mg(2+)/Ca(2+) antagonism as activators of the membrane ATPase was not directly related to Mg(2+) binding by the membranes. The efficiency of the ATPase release from M. lysodeikticus membrane by 3 mM tris(hydroxymethyl)aminomethane-hydrochloride buffer (pH 7.5) was inversely proportional to the concentration of external and/or bound Mg(2+). Deoxycholate (DOC) (1%) solubilized the ATPase from all types of membrane. All the soluble ATPases behaved as Ca(2+)-ATPases, but the DOC-soluble fractions showed degrees of latency like those of the original membranes. The DOC-soluble ATPase preparation revealed a vesicular structure and complex protein patterns by sodium dodecyl sulfate gel electrophoresis. We propose that ATPase latency is modulated via a Mg(2+)-ATPase-membrane complex.