The "Best Estimate Plus Uncertainty" (BEPU) approach has been gaining acceptance in the nuclear safety community, i.e. among regulators, policy makers and utilities. In its framework, one of the major sources of uncertainty considered is the experimental uncertainty. The paper first illustrates the importance of this type of uncertainty in the overall BEPU process. The different levels of maturity of the computational tools present different types of challenges for the generation of experimental data and their associated uncertainties .The paper presents a review the best practices adopted by the field of neutron transport to carry out measurements and determine reliable experimental uncertainties. An experiment must not be resumed to the measurement phase and to the choice of instrumentation, it includes some crucial phases, such as the conversion of raw data into the physical data which might introduce bias, i.e. the so-called Ouroboros paradigm. Finally, the rigorous approach used in the framework of international benchmarking collaborations in the field of neutron transport is summarized for the accurate reporting of experimental data and associated uncertainties. The detailed guidelines remain applicable to the other fields of nuclear engineering even though some of the steps (sensitivity analysis and combination of uncertainties) may be very difficult to perform in the fields of thermal-hydraulics and fuel performance where non-linear problems featuring important parameters with non-Gaussian distribution can be encountered.