Background: TKIs are standard treatment for non-small cell lung cancer (NSCLC) which harbors an actionable mutation. TKIs have a unique toxicity profile, distinct from conventional chemotherapy. Although the manufacturer of each TKI often issues a specific safety precaution on its label, no set schedule has been widely adopted. We surveyed the safety monitoring practice at a large academic institution. Methods: Electronic medical records of patients with ROS1, ALK or EGFR-mutated NSCLC who began a TKI during 2017-2021 were retrospectively reviewed. For each patient, the observation period started from the date of TKI prescription and lasted through day 60. Safety metrics were formulated based on drug label recommendations of each TKI. For each treatment course, the occurrence of 2 index safety monitoring activities (Table) were recorded. Results: Analysis included 130 treatment courses: 82 osimertinib, 20 alectinib, 15 crizotinib and 13 lorlatinib. Median age was 64.3 years, with 66% female, 81% white, 91% ECOG 0/1, 28% zero comorbidity, and 99% non-smoker. For osimertinib, EKG was obtained at baseline (within 30 days of treatment initiation) in 69.5% of observations, and LVEF, in 26.8%. By 60 days, these numbers barely increased (Table). Overall, absence of any index safety monitoring activity occurred in 41 of 130 (32%) treatment courses. Multivariable logistic regression analysis found a trend toward less monitoring among non-white patients: Odds ratio 0.35 (95% CI 0.11-1.09, p = 0.07). No association was found between the number of cardiac risk factors and LVEF monitoring or the use of QT prolonging medication and QTc monitoring. Conclusions: We observed a wide variation in the safety monitoring practice during TKI initiation. Deviation from the manufacturer recommendation was most pronounced for CPK monitoring in alectinib. Future study is needed to define the clinical significance of deviation and to identify barriers to safety monitoring.