Leaf area index (LAI) is a variable of primary importance for crop monitoring and it is usually measured by labor intensive destructive field sampling. Indirect non-destructive estimates by optical instruments are a competitive alternative to direct methods for frequent measurements and over large areas. The work presented here evaluated the adequacy and the range of reliability of the LAI-2000 estimates for rice (Oryza sativa L.) by comparison with LAI measurements derived by destructive sampling. Field data were collected in 2004 for an Indica type variety grown under different levels of nitrogen fertilization. The comparison showed that LAI-2000 estimates and destructive LAI measurements are well correlated (R2 > 0.8) whereas the correlation decreases when LAI values are lower than one (R2 1, the estimates derived by discarding the wide angle reading (fifth ring) of the optical instrument are closer to the actual LAI values compared to the standard LAI-2000 outputs. The regression line is, in fact, not statistically different (P = 0.25) from the 1:1 line. In these application conditions, LAI-2000 estimates are correct within +-32%. Therefore, if the range of applicability is restricted to LAI values greater than one and the wide angle readings are discarded, LAI-2000 is an appropriate instrument for in situ LAI estimation in paddy rice fields.