Currently, megavoltage CT studies in most centres with tomotherapy units are performed prior to every treatment for patient set-up verification and position correction. However, daily imaging adds to the total treatment time, which may cause patient discomfort as well as results in increased imaging dose. In this study, four alternative megavoltage CT imaging protocols (images obtained: during the first five fractions, once per week, alternating fractions and daily on alternative weeks) were evaluated retrospectively using the daily position correction data for 42 patients with lung cancer. The additional uncertainty introduced by using a specific protocol with respect to the daily imaging, or residual uncertainty, was analysed on a patient and population bases. The impact of less frequent imaging schedules on treatment margin calculation was also analysed. Systematic deviations were reduced with increased imaging frequency, while random deviations were largely unaffected. Mean population systematic errors were small for all protocols evaluated. In the protocol showing the greatest error, the treatment margins necessary to accommodate residual errors were 1.2, 1.3 and 1.7 mm larger in the left–right, superior–inferior and anterior–posterior directions, respectively, compared with the margins calculated using the daily imaging data. The increased uncertainty because of the use of less frequent imaging protocols may be acceptable when compared with other sources of uncertainty in lung cancer cases, such as target volume delineation and motion because of respiration. Further work needs to be carried out to establish the impact of increased residual errors on dose distribution.