The task of energy management for hybrid electric vehicles (HEV) consists in providing the requested traction power by splitting the demand between electrical and mechanical energy sources. The best performance can be achieved if the power split can be computed over the whole trip at once, e.g. using Dynamic Programming (DP). However, actual travel conditions are prone to change with respect to the assumed ones, so that updates are needed. Recomputing the optimal power split for the whole remaining route via DP is hardly feasible due to the actual travel conditions, so approximate solutions are needed. In this paper, we propose an online-capable strategy for a parallel HEV based on the minimization of total losses and heuristics for engine on/off and gear actions. It yields a very good approximation of the DP solution with little computational burden.