The lack of comprehensive language proficiency assessment models recognizing the nonlinearity and dynamicity of individual differences, as well as the fundamental differences between paper- and computer-assisted language assessments (CALA) served as the rationale to conduct the present study. Drawing on dynamic systems theory, a Nonlinear Dynamic Individual-Centered Language Assessment (NDICLA) model was developed and tested for its applicability in the computer-assisted language learning context. To test the effectiveness of the model, a random assignment experiment was conducted among 73 Iranian EFL learners. Results show that language proficiency gains of the experimental group were statistically and practically greater than those of the control group. The experimental group held largely positive attitudes towards NDICLA assessment practices, in terms of engagement, competence, and performance. Lastly, the NDICLA model users perceived the model to be effective in terms of accessibility, content, and function. The integration of the computer-assisted language learning developments into a multifaceted CALA model is a significant step towards a dynamic, interactive, and adaptive assessment model, in response to the non-linear dynamic individual differences.