An accurate calculation of the nondipole anisotropy parameter γ in the photoionization of helium below the N = 2 threshold is presented. The calculated results are in fairly good agreement with the experimental results of Krässig et al (2002 Phys. Rev. Lett. 88 203002), but not as good as the accuracy of the calculation should have warranted. A careful examination of the possible causes for the observed discrepancies between theory and experiment seems to rule out any role either of the multipolar terms higher than the electric quadrupole, or of the singlet–triplet spin–orbit mixing. It is argued that such discrepancies might have an instrumental origin, due to the difficulty of measuring vanishingly small total cross sections σtot with the required accuracy. In such eventuality, it might be more appropriate to use a parameter other than γ, such as for instance the drag current, to measure the nondipole anisotropy of the photoelectron angular distribution.