The nitrogen fluxes from soil to plant population were investigated in 3 mountain grasslands on andepts at 1000-1350 m asl, selected for their contrasting levels of intensification. The mean nitrogen uptakes in the absence of mineral N fertilization and leguminous plants decrease with the number of years without nitrogen restitution (Fig. 1). The mean values were 29, 75 and 100 kg N.ha-1.an-1 for the poor grassland (P), the intensified grassland (M) and the grassland (R) respectively. The apparent efficiencies of the nitrogen fertilizer, measured in the 5th year, increased from 32.4 (P) to 26.1 (M) and 51.7 (R) (Tab. I). However, for all grasslands the nitrogen recoveries in the harvested parts conformed to the same exponential relation for the total available nitrogen supplied by the soil and by fertilization (Fig. 2). The real and apparent recoveries were measured on grasslands without previous exhaustion of N by cutting : the apparent recoveries were 63 (P) and 77% (M), and the real recoveries 25 (P) and 32% (M) (Tab. III). The apparent coefficients due to seasonal supply decreased with the date of supply : in the poor grasslands, annual efficiencies account for 39 and 12% respectively of the applications in June and July; in the intensified grassland, annual efficiencies account for 59, 30 and 7% respectively of the applications in May, June and August. The nitrogen uptake from the soil in the presence of mineral fertilization is 71 (P) and 133 kg N.-ha-1,year-1 (M) as measured by 15N, instead of 33 (P) and 65 (M) as measured in the absence of fertilization (Tab. IV). The discussion focuses on the possibility of a real added nitrogen interaction and the evaluation of the in situ gross mineralization.