In a mass balance of 15N-labelled nitrate added to soil grown with pea or barley, denitrification estimates using the acetylene-inhibition technique were compared with unaccounted for 15N. During the growth season of 1989, which was drier than average, N losses due to denitrification estimated by the acetylene-inhibition technique were negligible. A substantial amount of fertilizer N was unaccounted for by the 15N mass balance, especially in the pea plots. The loss took place during the period of grain-filling in which no leaching occurred, and was accompanied by a decrease in 15N content of the plants. Volatilization of ammonia from the aerial parts of the plants is a possible explanation of the observed loss. An estimation of denitrification relying only on the 15N mass balance would have resulted in an overestimation of denitrification.