ABSTRACT Conceptually, sports-specific training should not influence measures of impairment used to classify Para athletes. This study evaluated the extent to which measures of strength, range of movement and coordination developed for Para swimming classification changed in response to a performance-focused swimming programme. A five-phase multiple-baseline, single-case experimental research design was utilized. Three participants with cerebral palsy and high support needs completed the 64-week study, which included two 16-week performance-focused swimming training blocks. Swimming speed, isometric shoulder extension strength, shoulder flexion range of movement and upper limb coordination were monitored throughout. Interrupted Time-Series Simulation Method analysis demonstrated large, significant changes in swimming speed (m/s) during the first (d = 2.17; 95% CI 0.45–3.88; p = 0.01) and second (d = 2.59; 95% CI 1.66–3.52; p = 0.00) training blocks. In contrast, changes in strength, range of movement and coordination were predominantly trivial and non-significant. This was the first study to investigate training responsiveness of measures developed for Para sport classification. Results indicate that despite significantly improved swimming performance, impairment measures remained relatively stable, and therefore these measures of impairment may be valid for the purposes of Para swimming classification. Further research is required in elite athletes, different sports and different impairment types.