This study proposes to develop a continuous, high-efficiency laboratory fibre fractionator capable of producing large quantities of well-fractionated fibres. The concept requires modifying a conventional Bauer-McNett classifier to include feedback and continuous feed flows. A computer model of the proposed fractionator was first constructed using Simulink in MATLAB. Simulations were performed to study effects of six system parameters on two fractionation characteristics. Results indicated that while large screen apertures led to higher cut-off length, they also resulted in less distinct separation. Simulations were validated with experimental tests performed on a prototype fractionator. Excellent agreement was found between experiment and simulation.