The characterization of the incident photon beam is usually divided into its dependence on collimator setting (head-scatter factor) and off-axis position (primary off-axis ratio). These parameters are normally measured "in air" with a build-up cap thick enough to generate full dose build-up at the depth of dose maximum. In order to prevent any influence from contaminating electrons, it has been recommended that head-scatter measurements are carried out using a mini-phantom rather than a conventional build-up cap. Due to the volume of the mini-phantom, the effects from attenuation and scatter are not negligible. In relative head-scatter measurements these effects cancel and the head scatter is thus a good representation of the variation of the incident photon beam with collimator setting. However, in off-axis measurements, attenuation and scatter conditions vary due to beam softening and do not cancel in the calculation of the primary off-axis ratio. The purpose of the present work was to estimate the effects from attenuation and phantom scatter in order to determine their influence on primary off-axis ratio measurements. We have characterized the off-axis beam-softening effect by means of narrow-beam transmission measurements to obtain the effective attenuation coefficient as a function of off-axis position. We then used a semi-analytical expression for the phantom-scatter calculation that depends solely on this attenuation coefficient. The derived formalism for relative "in air" measurements using a mini-phantom is clear and consistent, which enables the user to separately calculate the effects from scatter and attenuation. For the investigated beam qualities, 6 and 18 MV, our results indicate that the effects from attenuation and scatter in the mini-phantom nearly cancel (the combined effect is less than 1%) within 12.5 cm from the central beam axis. Thus, no correction is needed when the primary off-axis ratio is measured with a mini-phantom.