We have investigated the microsecond rotational motions of the Ca-ATPase in rabbit skeletal sarcoplasmic reticulum (SR), by measuring the time-resolved phosphorescence anisotropy of erythrosin 5-isothiocyanate (ERITC) covalently and specifically attached to the enzyme. Over a wide range of solvent conditions and temperatures, the phosphorescence anisotropy decay was best fit by a sum of three exponentials plus a constant term. At 4 °C, the rotational correlation times were ∅1 = 13 ± 3 μs, ∅2 = 77 ± 11 μs, and ∅3 = 314 ± 23 μs. Increasing the solution viscosity with glycerol caused very little effect on the correlation times, while decreasing the lipid viscosity with diethyl ether decreased the correlation times substantially, indicating that the decay corresponds to rotation of the protein within the membrane, not to vesicle tumbling. The normalized residual anisotropy (A∞) is insensitive to viscosity and temperature changes, supporting the model of uniaxial rotation of the protein about the membrane normal. The value of A∞ (0.20 ±.02) indicates that each of the three decay components can be analyzed as a separate rotational species, with the preexponential factor Ai equal to 1.25✕ the mole fraction. An empirically accurate measurement of the membrane lipid viscosity was obtained, permitting a theoretical analysis of the correlation times in terms of the sizes of the rotating species. At 4 °C, the dominant correlation time (∅3) is too large for a Ca-ATPase monomer, strongly suggesting that the enzyme is primarily aggregated (oligomeric). From 4 to 20 °C, A3 decreases markedly (0.43 to 0.15) while A2 increases (0.14 to 0.40), suggesting a transition to a more mobile (less aggregated) species. At 20 °C, the dominant correlation time (∅2) has a value most consistent with a Ca-ATPase dimer, but a monomer is also possible. Above 20 °C, A2 decreases in favor of A1 which has a correlation time (∅1) most consistent with a monomer. Arrhenius plots of protein rotational rates (inverse correlation times) are linear, indicating no abrupt change in protein shape or lipid viscosity. In contrast, the van’t Hoff plots of the apparent protein-protein dissociation constants (A2/A3 and A1/A2) exhibit dramatic slope changes in the range of 18–23 °C, correlating well with changes in the energetics of Ca-ATPase activity and its proposed protein conformational transition. We propose that ATPase monomers and oligomers are in a temperature-dependent equilibrium, with monomers and dimers predominating above 20 °C and higher oligomers predominating at lower temperatures. These results support the hypothesis that protein-protein interaction plays an important role in determining Ca-ATPase activity.