Background <p>Recently, several classification systems have been proposed, among which the coronal plane alignment of the knee (CPAK) classification has gained widespread adoption. Nevertheless, existing calculation methods for CPAK do not account for the joint line convergence angle (JLCA), potentially compromising classification accuracy.</p> Methods <p>A modified CPAK classification incorporating JLCA was developed using the mechanical hip-knee-ankle angle (mHKA) and actual joint line obliquity (aJLO). A retrospective review was conducted on 120 patients (240 knees) with no or mild knee osteoarthritis (OA). Radiographic parameters—including mHKA, medial proximal tibial angle (MPTA), lateral distal femoral angle (LDFA), JLCA, and mechanical joint line angle (MJLA)—were measured, and arithmetic HKA (aHKA) and CPAK JLO were calculated accordingly. Knee alignment, joint line apex positions, and phenotypic classifications were compared between the conventional and the modified CPAK.</p> Results <p>The mean JLCA was 1.0° ± 1.1°, with only 46.7% of cases falling within the range of 0° ± 1°. The mean mHKA was 178.4° ± 3.3°, while the aHKA was − 0.5° ± 3.3°. The mean aJLO was 3.2° ± 1.7°, and the CPAK JLO was 173.1° ± 3.4°. Concordance between conventional CPAK and modified CPAK in terms of knee alignment was 75.4%, joint line apex position was 64.6%, and phenotypes was 51.7%. The most prevalent CPAK types were Type II (<i>n</i> = 109; 45.4%), followed by Type I (<i>n</i> = 57; 23.8%) and Type III (<i>n</i> = 51; 21.3%). In contrast, the most common modified CPAK types were Type II (<i>n</i> = 73; 30.4%), Type IV (<i>n</i> = 50; 20.8%), and Type V (<i>n</i> = 50; 20.8%).</p> Conclusion <p>The CPAK classification remains clinically relevant in knee arthroplasty, incorporating JLCA into CPAK evaluation is essential to improve classification precision and ensure more accurate representation of native knee.</p>