Integrated magnetic resonance (MR) imaging and radiotherapy (RT) delivery machines are currently being developed, with some already in clinical use. It is anticipated that the strong magnetic field used in some MR-RT designs will have a significant impact on routine measurements of dose in the MR-linac performed using ionization chambers, which provide traceability back to a primary standard definition of dose. In particular, the presence of small air gaps around ionization chambers may introduce unacceptably high uncertainty into these measurements. In this study, we investigate and quantify the variation attributable to air gaps for several routinely-used cylindrical ionization chambers in a magnetic field, as well as the effect of the magnetic field alone on the response of the chambers. The measurements were performed in a Co-60 beam, while the ionization chambers were positioned in custom-made Perspex phantoms between the poles of an electromagnet, which was capable of generating magnetic fields of up to 2 T field strength, although measurements were focused around 1.5 T. When an asymmetric air gap was rotated at cardinal angles around the ionization chambers investigated here, variation of up to 8.5 ± 0.2 percentage points (PTW 31006 chamber) was observed in an applied magnetic field of 1.5 T. The minimum peak-to-peak variation was 1.1 ± 0.1% (Exradin A1SL). When the same experiment was performed with a well-defined air gap of known position using the PTW 30013 chamber, a variation of 3.8 ± 0.2% was observed. When water was added to the phantom cavity to eliminate all air gaps, the variation for the PTW 30013 was reduced to 0.2 ± 0.01%.