Deep muscle training has become the focus of research and exercise for patients with chronic neck pain. The objective of this in vivo study was to establish a non-invasive assessment tool for the activation of deep cervical muscles. The pattern of the change in the thickness of the cervical multifidus is described with a mathematical equation and used to compare the changes among different levels of resistance (0%, 25%, 50%, 75%, and 100%) and at different cervical levels (fourth, fifth, and sixth cervical (C4, C5, and C6) vertebrae). Twenty asymptomatic subjects (five women and 15 men; 24.3+/-4.7 years old) were recruited for this experiment. Ultrasonography (US) with synchronized force recording was used to measure the thickness of the cervical multifidus during progressive isometric extension against resistance. Linear and quadratic models were used to estimate the patterns of change in the thickness of cervical multifidus in relation to force. Two-way analysis of variance with repeated measurement and post hoc analysis were used to investigate the differences in thickness. The change in thickness and force was better fitted by quadratic model (y=ax(2)+bx+c) than by the linear model. The thickness at 50% of maximum contraction was significantly increased compared with that at 25% of maximum contraction. This quantitative non-invasive measurement may provide an assessment tool for further investigation for the physiological function of the deep muscles. Further research is required to investigate whether the change of thickness was predominately determined by the recruitment of muscle fibers or the extensibility of non-contractile tissues.