Electron storage rings of synchrotron light sources are typically filled only for several minutes day(-1). During these injections, the dose rates outside the shielding walls can be quite high even if the annual dose personal dose equivalent is below 1 mSv a(-1), which is the case at most synchrotron light sources. During the injection process, there is a time structure of short pulses (often defined by the convolution time of the synchrotron) having a pulse length of several 100 ns and a repetition rate between 1 and 10 Hz, which is defined by the acceleration processes in synchrotrons or linacs. Under these conditions, high measurement errors are possible, especially for neutron monitors that are based on proportional counters, which is the case for most commercially available neutron detectors. In this article, the authors' investigations of different neutron monitors are presented and how these results depend on the beam parameters, neutron fields and detector properties is shown. The experiments were conducted at the synchrotron light sources BESSYII and Metrology Light Source with pulsed electron beams of 1.7 GeV and 100 MeV, respectively. Other experiments were conducted at the HZB cyclotron with a 68-MeV pulsed proton beam that hits a spallation target. Fluka calculations of dose rates and neutron spectra were performed for the experiments. Correction formulas that are valid even in the saturation range were derived.