Abstract Ground penetrating radar (GPR) is a popular technique for imaging and interpreting sedimentary architecture. However, current literature shows a wide range in the quality of information provided on the GPR methodology and processing technique. It is therefore difficult to judge the validity of the GPR interpretations and this produces inherent difficulties for comparison between surveys. This paper describes the key steps required to collect, process and interpret GPR surveys in sandy fluvial sediments. GPR data from the South Saskatchewan River, Canada, are used to illustrate each stage of data collection and processing. Particular attention is given to the appropriate set-up conditions for the GPR software and hardware, the selection of data-processing techniques and velocity analysis. Methods for the interpretation of GPR reflectors are also investigated using ground-truth control provided by a cut-face exposure. This paper presents recommendations for a systematic and rigorous methodology for the collection, processing and interpretation of GPR data in sandy fluvial environments. The paper suggests that all data-collection parameters and processing steps should be recorded or tabulated in any GPR publication to facilitate comparisons between surveys.