Spasticity of the elbow was generally assessed by repeated passive stretch movement, including the modified Ashworth Scale (MAS) from physiotherapist, and biomechanics analysis of the movement. The MAS-based method depends on the subjective evaluations and the performance of biomechanics analysis assessment is affected by the individual difference. Therefore, the normalization to reduce the individual difference for the assessment of spasticity is very important. In this study, the elbow spasticity was assessed with MAS by one skillful physiotherapist and biomechanics measurements during repetitive passive isokinetic movements at velocity of 60 degree$/$second. 20 post-stroke patients with elbow spasticity caused by hemorrhagic cerebral damage were divided into three groups according to the MAS grades (MAS $=1, 1+$, 2). The torque and position were recorded when the patients extension their elbows passively. The mean stiffness and the mean torque features of the passive isokinetic were calculated. Two normalization factors for biomechanics analysis assessment were investigated: body weight normalization factor and maximum isometrics volunteer contraction normalization factor. Spearman correlation analysis was used to investigate the relationship between the features and spasticity grades. The results showed that the correlation between MAS and two biomechanics features (mean stiffness, mean torque) were significant improved. For mean stiffness feature, the correlation coefficients were $-0.313, -0.563$ and -0.603 individually for non-normalization, body weight normalization and maximum isometrics volunteer contraction normalization. For mean torque feature, the correlation coefficients were $-0.260, -0.523$ and -0.691, respectively. These results suggest that the normalization methods would be helpful for the assessment of spasticity in biomechanics and will be a necessary way of spasticity estimation in clinical methods.