Prompted by environmental and agronomic concerns, emissions of nitrous oxide (N20) and nitrogen (N2) gases from soils under sugar cane were monitored by the acetylene inhibition technique at 4 sites fertilized with and without 140 kg N ha-1. The data obtained showed that the rates of N20IN2 emission from the soils cannot be predicted by any single soil variable. Moreover in spite of site to site variabilities. the temporal pattern of the gaseous N emissions invariably showed a base level emission rate ranging from IO to 50 g N ha-I d-1. Upon that background flux were pulses of emission which could be as high as 5 times the baseline flux. Following N fertilizer application, large pulses of gaseous N escape with the flux attaining a peak of 500 g N ha- I d-1 were only observed in the superhumid regions (> 3 500 mm rain y-') such as Belle Rive and Union Park. At the 2 sites (Reduit and Parnplemousses) in the low rainfall areas (< I 500 mm rain y·'). the rates of gaseous N loss from soils were hardly altered by fertilizer application. The results therefore showed that gaseous losses of fertilizer N could be of agronomic significance only in the high rainfall zone. On average 15% of the 140 kg N ha-I applied at Belle Rive escaped to the atmosphere over a sugar cane growing season. Soil wetness further controlled the ratio N20:N2 emitted from the soils. Conversion of N20 to N2 gas was of significance at Belle Rive and Union Park where more than 80% of the soil pore space was shown to be water-filled for frequent prolonged periods. Though the present study showed that gaseous losses of fertilizer N may in general not be of agronomic or economic relevance. The magnitude of yearly N20 emissions from sugar cane fields which ranged from 5 to I7 kg N ha-1, may be of concern from the environment point of view.