Light-use efficiency （LUE） is a critical parameter in many primary production models for estimating ecosystem carbon exchange. The application of these models on regional and global scale is restricted because of the difficulty of retrieving LUE from airborne and satellite remote sensing images. Vegetation chlorophyll fluorescence is a direct indicator of plant physiology. In this paper, a diurnal experiment was carried on maize on July 5, 2008. The canopy radiance spectra and tower-based flux data were acquired synchronously to test the possibility of retrieving LUE by the solar-induced vegetation ChlF signals. The canopy net primary production （NEP） values were calculated using eddy covariance measurement by a CSAT3-Li7500 Flux system, and the gross primary production （GPP） was also calculated by adding the simulated day time respiration. Two kinds of LUE based on GPP （LUEGPP） and NEP （LUENEP） were defined by dividing the absorbed photosynthetic active radiation （APAR）. The ChlF signals at 760nm and 688nm were also separated from the reflected radiance spectra based on Fraunhofer line depth algorithm in the two oxygen absorption bands. The ChlF signals were strongly correlated with photosynthetic active radiation （PAR）, especially the ChlF at 760nm （R2>0.99）. Both NEP and GPP had a significant correlation with ChlF. Furthermore, LUEGPP was negatively correlated with the ChlF’s relative intensity at 688nm and 760nm, with a correlation coefficient R2 of 0.6331 and 0.7861 respectively. Moreover, the LUE models based on the solar-induced vegetation ChlF signals were compared to some popular vegetation Indices （VIs） from the canopy reflected spectra. Canopy LUEGPP was proved able to be estimated from the remotely sensed ChlF signals.