Potassium fertilization recommendations for alfalfa (Medicago sativa L.) vary depending on the area of production, soil levels, and crop management. Th e objectives of this study were to determine the impact of K fertilization on irrigated alfalfa yield, yield components, and soil exchangeable K (K e ) values in a Mediterranean climate. A fi eld experiment was conducted during a period of 4 yr (2002)(2003)(2004)(2005)(2006) in Spain, on a soil with moderate levels of K e (161 mg K kg -1 ). Th e treatments applied were fi ve annual rates of K (0, 100, 200, 300, and 400 kg K ha -1 ). Total 4-yr dry matter (DM) yields averaged 74.9 Mg ha -1 without diff erences between K fertilizer rates, although the unfertilized control produced the lowest DM yields. Th e 4-yr crop uptake of K reached 1738 kg ha -1 with the application of 400 kg K ha -1 yr -1 , and 756 kg K ha -1 for the 0 K fertilization. Th e removal was greater than the K applied. Th e results suggest that large amounts of K should not be applied to alfalfa because the crop uses excess K without increasing yield. Soil K e concentrations decreased every year for all K rates except for the highest treatment of 400 kg K ha -1 yr -1 . Lack of K fertilization did not aff ect stand density but the shoot weight was the yield component most closely related to K fertilization. * Signifi cant at the 0.05 level; ns, not signifi cant. ** Signifi cant at the 0.01 level. † Year and K rate main effects and the Year × K rate interaction for the period (2002-2005) were signifi cant at the 0.01 level. ‡ LSD(0.05) value for the year effect is 0.55 Mg ha -1 . * Signifi cant at the 0.05 level; ns, not signifi cant. ** Signifi cant at the 0.01 level. † Year and K rate main effects and the Year × K rate interaction for the period 2003-2005 were signifi cant at the 0.01 level. ‡ LSD(0.05) for year effect on K concentration is 1.0 g kg -1 and for K removal is 23 kg ha -1 .