Forty-four soil clays and four reference phyllosilicate clays were used to establish qualitative and quantitative relationships between clay mineral characteristics and K fixation. The soil clays were obtained from Vertisols, Vertic Mollisols, Haploxerolls, Xerochrepts, and a Palexeralf, from three semiarid regions of northwestern Morocco. The soil clays contained beidellite, montmorillonite, vermiculite, and illite in widely varied abundances. Both total and tetrahedral cation-exchange capacity (CEC) were measured. Octahedral CEC was calculated by difference. Tetrahedral CEC was determined after saturation with Li and heating at 300-degrees-C to reduce octahedral charge to near zero. Potassium fixation was determined in both wet (1:4 soil/solution, 16 h on a shaker) and dry conditions (after drying for 24 h at 70-degrees-C) using three different levels of added K. Total CEC was well correlated with tetrahedral CEC (r2 = 0.76, P < 0.001). About half of the smectitic and vermiculitic soil clays have > 50% of their total charge due to tetrahedral charge. The amount of K fixed increased with added K, whereas the percent K fixed relative to total added K decreased. The K fixation by the clays studied was best correlated with total CEC for a given amount of K added. It was also well correlated with tetrahedral CEC. Simple regression analysis did not show any significant effect of octahedral CEC. A multiple regression analysis, however, demonstrated that both components of total CEC, tetrahedral and octahedral, have a simultaneous effect on K fixation. Tetrahedral CEC explained more of the variation (64%) than octahedral CEC (36%). This greater effect was attributed mainly to the proximity of tetrahedral charge to the interlayer space of 2:1 phyllosilicate clays.