Sun-induced chlorophyll fluorescence (Fs) is the radiation flux emitted from chlorophyll molecules and can be used as a remote sensing (RS) observable to be linked to plant photosynthesis. Recently, significant progress has been made to quantify Fs from RS data, but both retrieval and interpretation of Fs remain challenging. In the case of airborne sensors with a medium spectral resolution (b2–4 nm), Fs is typically estimated using the Fraunhofer Line Depth (FLD) approach focusing on atmospheric O2 absorption bands. Most critical for accurate Fs retrievals based on such methods is the characterization of atmospheric scattering and absorption processes during data acquisition. So far, detailed experimental evidence on the retrieval accuracy of airborne measured Fs is lacking. We performed an experiment using a low-flying aircraft equipped with a non-imaging spectrometer acquiring medium spectral resolution data during the course of one day, using a repeat-track approach with changing flight altitudes. Fs in the near infrared was retrieved using a semi-empirical approach constraining the FLD based Fs retrieval from the O2-A absorption band at 760 nm by using non-fluorescent surfaces. We used a local sensitivity analysis to assess Fs retrieval biases related to observational and atmospheric parameters. Our results demonstrate a reliable Fs retrieval from airborne data using reference surfaces and indicate the need for accurate knowledge of atmospheric scattering and absorption processes. This study contributes to an estimation of the total error budget of Fs retrievals and will serve as a practical guideline for Fs retrieval schemes to be applied to medium resolution airborne spectroscopy data.