In this paper, we will discuss the prospects of deducing the proton-air cross section from fluorescence telescope measurements of extensive air showers. As it is not possible to observe the point of first interaction X1 directly, other observables closely linked to X1 must be inferred from the longitudinal profiles. This introduces a dependence on the models used to describe the shower development. The most straightforward candidate for a good correlation to X1 is the depth of shower maximum Xmax. We will discuss the sensitivity of an Xmax-based analysis on σp−air and quantify the systematic uncertainties arising from the model dependence, parameters of the reconstruction method itself and a possible non-proton contamination of the selected shower sample.