In Plug-in hybrid electric vehicles (PHEVs) typically combine several power sources, which are coordinated by means of an optimal energy management strategy. When considering the so-called blended mode, in which the engine is regularly used over a trip, the shape of battery state-of-charge (SoC) trajectory over travelled distance is of particular importance for achieving minimum fuel consumption. The paper deals with in-depth analysis of optimal SoC trajectories obtained by off-line control variable optimization of a PHEV-type city bus given in parallel (P2) powertrain configuration. The optimization is conducted by using a dynamic programming-based optimization algorithm for a wide range of driving cycles and operating scenarios. It is found that, as opposed to usually assumed linear-like near-optimal shape, the SoC vs. travelled distance trajectory can take on significantly different optimal shapes for non-zero road grade profiles or driving cycle with relatively long distance. The emphasis is on analyzing root causes for such behavior and its implications to fuel consumption