We present an experimental and theoretical study of the pressure dependence of the Knight shift of 23 Na and 133 Cs in sodium and cesium metal, respectively. The sodium shift has been measured, employing the diamondanvil cell technique, up to about 8 GPa, and our previous discovery of a shift minimum around 1.5 GPa has been confirmed. The temperature dependence of the shift results solely from thermal expansion. The cesium shift, at 295 K, increases by 74% between normal pressure and 2.1 GPa. The theoretical studies of the sodium shift are based on a self-consistent band structure calculation with the scalar-relativistic linear muffin tin orbital method and the local density approximation. Using an appropriate description of the volume dependence of the hyperfine field, our calculations lead to a correct prediction of the Knight shift minimum. Differences between spin-restricted and exchange-enhanced calculations are discussed.