The increased concern about cost and potential environmental impact of fert
ilizers emphasizes the importance of improving the accuracy and soil specif
icity of fertilizer recommendations. Buffering capacity is considered to be
a key factor in understanding phosphorus (P) availability in the soil and
crop response because it controls the rate at which P is supplied or deplet
ed from the soil solution. Phosphorus buffering indices were determined fi
om sorption isotherms developed using 18 soils from the Abda, Chaouia, and
Ben Sliman arid and semiarid zones of Morocco. Soil characteristics were al
so used to develop prediction equations of these buffer indices. The greenh
ouse study was also conducted where wheat (Triticium aestivum:) was grown i
n 13 soils under four P rates (0, 3.4, 6.7, and 13.4 mg P kg(-1) soil). Cri
tical soil test P levels for wheat under greenhouse conditions were determi
ned. Maximum buffering capacity (MBC) of individual soils was incorporated
into the Mitscherlich equation which was modified to determine the P fertil
izer requirement. Our soils showed a contrasting ability to adsorb P. Maxim
um P adsorption (Xm) varied from 146 to 808 mg P kg(-1) soil. Averaged acro
ss regions, soils from Chaouia adsorbed more P at maximum adsorption as com
pared to the Abda and Ben Sliman soils, suggesting that the soil from each
region has to have specific P recommendations. Maximum buffering capacities
ranged from 35 to 404 mg P kg(-1). Differences between soils to react with
P was also demonstrated in a kinetic study we conducted. The inclusion of
MBC in fertilizer P recommendations did not significantly increase the prec
ision of current applied P at infinity 0.05. However, calculations of P req
uirement, assuming the soil test in all of our 13 soils was 3 mg P kg(-1) s
howed that the P fertilizer required to achieve 90% of the maximum yield va
ried from I to 15 mg P kg(-1) depending on the MBC of each soil. However, t
he fertilizer P recommendation determined by the usual method was 12 mg P k
g(-1) for all soils with a soil test of 3 mg P kg(-1). These results show t
hat incorporation of buffer indices into a P requirement model should incre
ase the accuracy of P recommendations and overcome the problem of over-fert
ilization in sandy soils and under-fertilization in clay soils.