To assess the sodium MRI signal loss resulting from typically used RF excitation pulses in human skeletal muscle, patellar cartilage, and skin. A double flip-angle experiment was performed 3 times on the knees of 5 healthy volunteers with prescribed ω1 = γB1 of 1.67 kHz, 0.333 kHz, and 0.167 kHz. This was done to search for ω1 -dependent increased rates of sodium-23 central resonance flipping known to result from residual quadrupole splitting (ωQ ), as this flip-angle effect is associated with signal loss. This study facilitated in vivo regression of Gaussian-distributed residual quadrupole splitting SD (ωQ(SD) ) as well as T2fast and T2slow . Signal loss predicted from simulation was then compared with images acquired using 90° RF pulse lengths of 0.5 ms, 0.25 ms, and 0.15 ms. Sodium-23 central resonance flipping was significantly greater than prescribed (44% cartilage, 23% skin, 9% muscle) using ω1 = 0.167 kHz, but only 4% cartilage, 5% skin, and 2% muscle using ω1 = 1.67 kHz. Regression yielded ωQ(SD) = 420 ± 50 Hz for cartilage but no significant ωQ(SD) for skin or muscle. This points to rapid biexponential relaxation as the cause of the flip-angle effect for skin/muscle. The T2fast(60%) /T2slow(40%) values were 1.6 ± 0.8 ms/16.1 ± 2.5 ms for muscle, 2.7 ± 0.9 ms/18.4 ± 2.5 ms for cartilage, and 0.4 ± 0.1 ms/9.3 ± 1.7 ms for skin. Simulation predicted signal loss of 6% ± 3%, 3% ± 1%, and 2% ± 1% for muscle, 16% ± 3%, 6% ± 1%, and 3% ± 1% for cartilage, and 26% ± 7%, 15% ± 4%, and 10% ± 3% for skin when using 90° RF pulse lengths of 0.5 ms, 0.25 ms, and 0.15 ms, matching experiment. High-power (short) RF pulses are necessary to reduce excitation-related signal loss, particularly for sodium-23 imaging of cartilage and skin.