The functional neuroanatomy of executive function critically involves the dorsolateral prefrontal cortex. Transcranial direct current stimulation (tDCS) has been established as a noninvasive tool for transient modulation of cortical function. Here, we examined the effects of tDCS of the left dorsolateral prefrontal cortex on planning function by using the Tower of London task to evaluate performance during and after anodal, cathodal (1 mA, 15 min), and sham tDCS in 24 healthy volunteers. The key finding was a double dissociation of polarity and training phase: improved performance was found with cathodal tDCS applied during acquisition and early consolidation, when preceding anodal tDCS, but not in the later training session. In contrast, anodal tDCS enhanced performance when applied in the later sessions following cathodal tDCS. Our results indicate that both anodal and cathodal tDCS can improve planning performance as quantified by the Tower of London test. Most importantly, these data demonstrate training-phase-specific effects of tDCS. We propose that excitability decreasing cathodal tDCS mediates its early beneficial effect through noise reduction of neuronal activity, whereas a further adaptive configuration of specific neuronal connections is supported by excitability enhancing anodal tDCS in the later training phase by enhanced efficacy of active connections. This gain of function was sustained in a follow-up 6 and 12 months after training. In conclusion, the specific coupling of stimulation and training phase interventions may support the treatment of cognitive disorders involving frontal lobe functions.