Five psychological refractory period (PRP) experiments were conducted with an especially time-consuming first task (Experiments 1, 3, and 5: mental rotation; Experiments 2 and 4: memory scanning) and with equal emphasis on the first task and on the second (left-right tone judgment). The standard design with varying stimulus onset asynchronies (SOAs) was extended by a condition with blocked SOAs in Experiments 3 and 4. Based on the optimization account (Miller, Ulrich, & Rolke, 2009) it was expected that participants would--at short, but not long SOAs--tend to process the relatively fast central stage of Task 2 before the time-consuming central stage of Task 1 and consequently emit the response to Task 2 before the response to Task 1. Such an optimization tendency was found, more so for the mental rotation task and for the blocked SOA condition. The results indicate that preparation, Task 1 characteristics, and TRT (total reaction time) optimization are--among others--factors influencing central processing order in PRP tasks. (Contains 4 footnotes and 15 figures.)