An experiment was carried out in a naturally ventilated greenhouse to study the relationship between banana (Musa sp.) plant transpiration (Tr) measured with load cells, reference crop evapotranspiration (ET o ) calculated with five widely used models (i.e. the Priestley-Taylor, FAO radiation, Hargreaves, FAO Penman and FAO Penman-Monteith models) and pan evaporation (E pan ) measured with a standard Chinese 20 cm pan. Microclimatic conditions were measured inside the greenhouse. Results show that vapor pressure deficit and air temperature had good linear correlations to banana Tr with R 2 of 0.67 and 0.62, respectively. Among the five models tested, banana Tr and ET o calculated with the FAO-Penman model yielded the highest determination coefficient (R 2 = 0.67), followed by the FAO-PM model (R 2 =0.63), the FAO radiation model (R 2 =0.52), the Hargreaves model (R 2 =0.49) and the Priestley-Taylor model (R 2 =0.47). Banana transpiration Tr vs. E pan yielded an R 2 of 0.83, which is higher than the five models tested. In conclusion, the 20 cm pan can be useful for estimating banana Tr in greenhouses.