A Round Robin test was set up to obtain a cross comparison of participant's methods regarding a typical pressure relief device (bursting disc) sizing problem. The problem in question was a 'gassy' runaway reaction system: decomposition of 40% w/w dicumyl peroxide in 2,2,4-trimethyl-1,3- pentanediol diisobutyrate. The objective of the Round Robin exercise was to highlight the disparities in the participant's approaches and identify barriers and research needs. Two broad methods were used: one was the implementation of the DIERS method; the other was to follow a similarity or scale-up approach. The scale-up approach consists experimentally determining a minimum safe vent area and scaling up maintaining the vent area/volume ratio. This approach is similar to the one adopted by the UN committee on the Transport of Dangerous Goods for vent sizing of peroxide tankers using a 10 litre vessel. Very different results in terms of vent area were obtained from the seven participants. Analysis of the vent sizing calculations based on calorimetric data showed that the experimental conditions and their interpretation can significantly influence the estimated gas production rate and therefore the final vent area. However, the calculation of the vented mass flux stays largely responsible for the differences in calculated vent area. The critical factor lies in the assumption of single-phase (gas only) venting or two-phase (liquid and gas) venting. The reasons for assuming single-phase venting, which gives a lower vent area, are unclear and may not be justified. The comparison showed that the calorimetric and the similarity approaches diverged significantly when two-phase only venting was assumed. The Round Robin test highlighted the need for more experimental and modelling work towards predicting the nature of the vented fluid at large scale, and in the use of adiabatic calorimetry to determine the maximum gas production rate. Large-scale tests would allow the comparison, the validation and the improvement of the calorimetric and similarity approaches.