The experiment was carried out in a heavy soil with restricted drainage from autumn 1990 to autumn 1992 in a locality near Parral (Vllth Region). A seed mixture of 3 and 9 kg/ha of white clover (Trifolium repens L.) variety Regal and tall fescue (Festuca arundinacea Schreb.) variety Manade was sown, respectively. Treatments corresponded to the factorial combinations of 0, 30, 60 and 90 kg/ha of P (triple superhosphate) and 0, 50 and 100 kg/ha of K (potassium muriate) plus one treatment with 15 and 100 kg/ha of P and K, respectively. The fertílizer applications were made at sowing and annualy in autumn (P) and spring (K). A randomized block design with three replications in 2 x 6 m plots was used. Sefore sowing 2,000 kg/ha of lime and 220 kg/ha of gypsum were applied. Available P and K in soil was measured at the beginning of the experiment and in April 1991 and 1993. The total and specific D.M. production of these swards was measured at each cut. The low initial levels of P and K in soil increased slightly when P and K application rates were increased. The D.M. yields increased from 3.3 to 6.0 and 7.6 to 10.7 ton/ha in the first and second seasons, respectively, when P application was increased from 0 to 90 kg/ha. The P response tended to be linear in 1990/91 and quadratic in 1991/92 with a maximum gradient between 15 and 30 kg/ha. There was a significant interaction P x K in 1990/91 and there was no effect of K in 1991/92.