Proton beam therapy has advantages in comparison to conventional photon radiotherapy due to the physical properties of proton beams (e.g. sharp distal fall off, adjustable range and modulation). In proton therapy, there is the possibility of sparing healthy tissue close to the target volume. This is especially important when tumours are located next to critical organs and while treating cancer in paediatric patients. On the other hand, the interactions of protons with matter result in the production of secondary radiation, mostly neutrons and gamma radiation, which deposit their energy at a distance from the target. The aim of this study was to compare the response of different passive dosimetry systems in mixed radiation field induced by proton pencil beam inside anthropomorphic phantoms representing 5 and 10 years old children. Doses were measured in different organs with thermoluminescent (MTS-7, MTS-6 and MCP-N), radiophotoluminescent (GD-352M and GD-302M), bubble and poly-allyl-diglycol carbonate (PADC) track detectors. Results show that RPL detectors are the less sensitive for neutrons than LiF TLDs and can be applied for in-phantom dosimetry of gamma component. Neutron doses determined using track detectors, bubble detectors and pairs of MTS-7/MTS-6 are consistent within the uncertainty range. This is the first study dealing with measurements on child anthropomorphic phantoms irradiated by a pencil scanning beam technique.