Human-cat interaction (HCI) in rescue settings is often inconsistent, driven by individual caregiver beliefs rather than standardised evidence-based practice. This study evaluates the efficacy of an educational intervention for shelter workers, incorporating evidence based HCI principles designed to promote cat comfort and reduce the risk of cat human-directed aggression. A pre-post study followed employees and volunteers across three timepoints: a pre-intervention baseline (T1, n = 47), 2 weeks post-intervention (T2, n = 47), and a six-month follow-up (T3, n = 19). To assess intervention efficacy, participants’ real-time HCI were filmed at each time point and blind-rated by cat behaviour experts based on the degree to which participants’ cat directed behaviour aligned with evidence based best practice HCI principles (either ‘closely’, ‘somewhat’ or ‘not at all’ aligned). Participants also completed self-reported attitude and practice questionnaires that were similarly designed to assess alignment with best practice HCI principles and approaches. Composite ‘attitude’ and ‘practice’ scores were generated with higher scores reflecting greater alignment with best practice HCI principles. Cumulative Link Mixed Models (CLMM) revealed very strong evidence that participants real-time HCI ratings improved post-intervention [LRχ2(2) = 37.07, p < 0.001], with these gains remaining stable at the six-month follow-up. Specifically, the probability of receiving a “closely aligned” HCI rating increased from 2% (±3%) at T1 to 25% (±16%) at T2 (p = 0.008), rising further to 37% (±27%) at T3 (p = 0.023). At baseline prior to training, higher participant attitude scores were significantly predictive of higher real-time HCI ratings. This association disappeared post-intervention as both HCI ratings and attitude scores increased across the population, effectively reducing score variance and suggesting the training “levelled the playing field” across participants regardless of their starting beliefs. Staff with shorter service lengths (under 2 years) were nearly twice as likely to achieve higher HCI ratings post intervention compared to longer serving staff members, suggesting training may be more effective when delivered within shorter post-recruitment windows. The educational intervention was effective in promoting positive and sustainable improvements in participant’s HCI, highlighting the benefits of evidence-based training delivered to shelter workers.