There is a need to know flow capacities of common bottlenecks of cycle storages because an increasing number of large cycle storages are being designed. Flow capacity values are not available from literature nor is a suitable method to determine those values. This research contributes to filling these research gaps, by developing a method to measure capacity, and applying it to case studies. The method also offers a way to calculate the standard deviation of the flow, as well as a confidence interval for the capacity. This is possible because the flow is normally distributed. Knowing the variation of the flow, predictions can be made on how often a queue will emerge. The essence of the method is to measure in an event-based manner, discard the time at which there was no queue before the bottleneck, and then divide the remaining measurements into intervals of equal duration and treat those intervals as individual capacity measurements. The method has shown to yield a capacity value that is independent of the averaging interval. The standard deviation however, does not always vary as independent data from a normal distribution would. It is hypothesised that the autocorrelation could be used to correct for that. 95% confidence intervals for capacity were found as follows; check-in: 21.2 ± 1.2 (σ=3.5), stairs: 24.78 ±0.60 (σ=1.02).