To stably form a toner image of high image quality by controlling image density while maintaining a satisfactory image forming condition.

Exposure energy E and the amplitude Vpp of a developing bias are set to reference values (step S1), and while successively changing and setting the weighted average voltage Vavg of the developing bias in its variable range, patch images are formed (steps S2-S5). The density of the patch images is detected (step S6) and the optimum value of the weighted average voltage Vavg is calculated on the basis of the detected result (step S9). When the density of any patch image does not reach target density or exceeds the target density (steps S7, S8), the amplitude Vpp of the developing bias is increased or reduced (steps S10, S11) and then the weighted average voltage Vavg is controlled again.