Abstract Our objectives were: (1) to measure how N 2O and NO emissions from a chronosequence of forested land converted to pastures in the humid tropics of Costa Rica had changed in 4 yr, and (2) to relate these emissions to indices of N availability. We observed lower mean N 2O (11.9 ng N cm −2 h −1) and NO (3.5 ng N cm −2 h −1) emissions from pastures in 1996 compared to 1992 (N 2O: 39.9 ng N cm −2 h −1; NO: 5.8 ng N cm −2 h −1). Even so, N 2O emissions in recently formed pastures (13.8 ng N cm −2 h −1) were still higher than previously measured emissions from forests (7.0 ng N cm −2 h −1). Indices of N cycling, such as net N mineralization, nitrification potential, and extractable soil nitrate, decreased with pasture age, which we attributed to a decrease in substrate availability. Denitrification enzyme activity did not change significantly with pasture age, indicating that denitrification occurs at least sporadically at all sites and the presence of denitrifying enzymes is not as strongly linked to N availability as is the presence of nitrifying enzymes. There were no significant correlations between N 2O and NO emissions and indices of N cycling. While this may indicate that the processes are not closely related, we believe that sampling of nitrogen oxide emissions in 1996 was inadvertently biased towards exceptionally dry soil conditions. This sampling bias limited the probability of observing large nitrogen oxide emissions associated with episodic denitrification. Results from chronosequence studies should be interpreted with caution especially for variables which depend on local weather conditions at time of measurement.