OBJECTIVES:  On October 1, 2015 payers in US required all medical claims to be submitted using ICD-10-CM instead of ICD-9-CM. Increased quantity and specificity of codes in ICD-10-CM will likely enhance misclassification in claims data during transition. This study describes real-world ICD-10-CM coding patterns for several diseases during ICD-10-CM transition. METHODS:  Adjudicated claims in Truven Health MarketScan®Commercial and Medicare Supplemental databases for adult patients were used. Patients with ICD-9-CM and ICD-10-CM codes spanning the transition through January 31, 2016 were selected for: lung cancer (LC), pleural mesothelioma (PM), Alzheimer’s disease (AD), and diabetes mellitus (DM). LC was identified using ICD-9-CM=162.2-162.9 and ICD-10-CM=C34.*; PM using ICD-9-CM=163.* and ICD-10-CM=C45.0; AD using ICD-9-CM=331.0 and ICD-10-CM=G30.0, G30.1, G30.8, and G30.9; and DM using ICD-9-CM=250.* and ICD-10-CM=E10.*-E14.*. RESULTS:  Clinically relevant ICD-10-CM associated with ICD-9-CM LC (in ≥1% of ICD-9-CM cohort and rate ratio ≥2x versus matched other cancer controls) were C34.*. Clinically relevant ICD-10-CM associated with ICD-9-CM PM were C34.*, C38.4, C45.0, C45.7, C45.9, C76.1, C80.0, and C80.1. Of patients receiving an AD ICD-9-CM 331.0, 66% received one of the 4 G30.* ICD-10-CM, other clinically relevant codes observed at a frequency of at ≥10% included dementia and altered mental status (F03.90, F02.80, F02.81, and R41.82). Depending on DM type, ICD-10-CM yielded high sensitivity (97.6%-100%) for type 2 diabetes (T2D), when additional age restriction and treatment criteria were used. Corresponding values for type 1 diabetes (T1D) were 77.4%-86.8%. Clinically relevant ICD-10-CM associated with ICD-9-CM codes were E10.9 and E10.65 for T1D, and E11.9 and E11.65 for T2D. CONCLUSIONS:  We were able to identify real-world coding patterns in claims data during the transition. As providers gain greater familiarity with ICD-10-CM, coding practice will likely evolve over time. Researchers should continue to look for similar descriptive or validation studies of cohort algorithms relevant to the study period.