This article investigates practical methods to assess and compare different candidate physical layers for free-space optical transmissions. The example of an OOK signal with a simple APD receiver is used as a guiding case study throughout the paper. In this particular configuration only the signal fluctuations (scintillations) induced by the propagation of the optical link through atmospheric turbulence impact the optical signal. The first part of the paper discusses the type of data upon which physical layer evaluation can be conducted. It describes experimental data obtained from field trial and synthetic time series obtained from atmospheric propagation channel simulators. In the second part of the paper, a practical methodology is introduced to evaluate off-line physical layer performance on experimental data records. The final part of the paper presents the application of the proposed methodology to the case study under consideration. Specifically, the receiver model is first introduced. Then, the performance of different forward-error correction (FEC) codes is evaluated on selected uncoded transmission experiments records.