OBJECTIVE: To identify biomechanical and clinical parameters that influence knee flexion (KF) angle at initial contact (IC) and during single limb stance phase of gait in children with spastic cerebral palsy (CP) who walk with flexed-knee gait. DESIGN: Retrospective analysis of gait kinematics and clinical data collected from 2010-2013. SETTING: Motion & Gait Analysis Laboratory at Lucile Packard Children's Hospital, Stanford, CA. PARTICIPANTS: Gait analysis data from persons with spastic CP (Gross Motor Function Classification System [GMFCS] I-III) who had no prior surgery were analyzed. Participants exhibiting KF ≥20° at IC were included; the more-involved limb was analyzed. METHODS: Outcome measures were analyzed with respect to clinical findings, including passive range of motion, Selective Motor Control Assessment for the Lower Extremity (SCALE), gait kinematics, and musculoskeletal models of muscle-tendon lengths during gait. MAIN OUTCOME MEASURES: KF at IC (KFIC) and minimum KF during single-limb support (KFSLS) were investigated. RESULTS: Thirty-four participants met the inclusion criteria, and their data were analyzed (20 males and 14 females, mean age 10.1 years, range 5-20 years). Mean KFIC was 34.4 ± 8.4 degrees and correlated with lower SCALE score (ρ = -0.530, P = .004), later peak KF during swing (ρ = 0.614, P < .001), and shorter maximal muscle length of the semimembranosus (ρ = -0.359, P = .037). Mean KFSLS was 18.7 ± 14.9 and correlated to KF contracture (ρ = 0.605, P < .001) and shorter maximal muscle length of the semimembranosus (ρ = -0.572, P < .001) and medial gastrocnemius (ρ = -0.386, P = .024). GMFCS correlated more strongly to KFIC (ρ = 0.502, P = .002) than to KFSLS (ρ = 0.371, P = .031). Linear regression found that both the SCALE score (P = .001) and delayed timing of peak KF during swing (P = .001) independently predicted KFIC. KF contracture (P = .026) and maximal length of the semimembranosus (P = .043) independently predicted KFSLS. CONCLUSION: Correlates of KFIC differed from those for KFSLS and suggest that impaired selective motor control and later timing of swing-phase KF influence knee position at IC, whereas KF contracture and muscle lengths influence minimal KF in single-limb support, findings with important treatment implications.