A proximal point trust region line search (PPTRLS) algorithm is proposed for the smooth unconstrained minimization problem. At each iteration, the algorithm solves approximately a quadratic subproblem by a truncated Newton method with step length control. We have established (i) a rule for updating the proximal parameter; (ii) a nonmonotone criterion for accepting the iterate; (iii) a related scheme for updating the quadratic model. The global convergence analysis is presented, together with comparative numerical results that validate the proposed approach.